Patients with brain metastases have previously had poor survival of only 3 to 4 months with non-surgical treatment.1 2 Substantial improvement has been achieved in recent years with the advance of systemic treatment and radiation techniques. Stereotactic radiosurgery (SRS) was first delivered with the Cobalt-60 Gamma Knife system by Leksell in 1951.3 Today, SRS can also be delivered via the linear accelerator (LINAC) system and proton beam system. It is usually indicated in patients with oligo-brain metastases (≤4) with a diameter of less than 4 cm.4 It is particularly advantageous for lesions in the deep brain parenchyma that are not easily accessible by surgery. A frame-based system was initially used to immobilise the patient. A frameless system was later developed to minimise patient suffering and was reported to have comparable outcomes to the framed-based system.5 Since the introduction of a frameless system, SRS or even fractionated stereotactic radiotherapy has been increasingly used to treat patients with oligo-brain metastases. Patients do not have to undergo painful frame placement. Rather, they undergo simple planning procedures over 2 consecutive days. The patient is required to return only for mould fitting and planning of computed tomography. Together with diagnostic fine-cut magnetic resonance imaging co-registration, oncologists can easily contour the target on the radiotherapy planning system. With the use of the ExacTrac system (Brainlab AG, Germany) to verify treatment position, the magnitude of error is reported to be only 0.7 mm, and the mean deviation between frame-based and image-guided initial positioning is just 1.0 mm (standard deviation, 0.5 mm).6 A frameless system became one of the choices of treatment in SRS and was included in the ASTRO policy.7 The recommended dosages according to the RTOG 9005 trial are 24 Gy, 18 Gy, and 15 Gy for tumours of ≤20 mm, 21-30 mm, and 31-40 mm in maximum diameter, respectively.8 For framed SRS, 1-year local progression-free survival (PFS) was reported to be up to 70% to 90%, and median overall survival (OS) of 6 to 12 months.9 10 11 12 13 14 15 The outcomes of frameless SRS have been reported only in limited series, with 1-year local control of 79% to 95%.5 16 17 18

Patient selection and tailor-made management are indeed challenging. Several scoring systems have been modelled to predict survival of patients with brain metastases, including the Radiation Therapy Oncology Group Recursive Partitioning Analysis (RTOG RPA),19 Basic Score for Brain Metastases (BSBM),20 the Score Index for Radiosurgery in Brain Metastases (SIR),21 Graded Prognostic Assessment (GPA),22 and Disease-Specific Graded Prognostic Assessment (DS-GPA).23 These scoring systems were developed at a time when treatment strategies were also rapidly evolving with the availability of more accurate diagnostic imaging, better radiotherapy techniques, and more effective systemic and targeted agents. A paradigm shift to more aggressive treatment of oligo-metastasis as a result of longer cancer survivorship now requires further validation of these scoring systems.

In this study, we reviewed the outcomes of patients who underwent LINAC-based frameless SRS and identified prognostic factors that affect survival. By doing so, we hope to gain a better understanding of which patients will benefit from SRS without jeopardising their quality of life.

Records of patients who underwent frameless SRS for limited brain metastases in a university-affiliated hospital between January 2010 and July 2015 were retrospectively reviewed. Patient data were extracted from paper records and the Clinical Management System of the Hospital Authority, Hong Kong by investigators in charge of the study. Data extracted included gender, age, type of primary malignancy, date of diagnosis of malignancy and brain metastases, extracranial disease status and control at treatment time, diagnostic and monitoring modalities, presence of convulsions, and steroid and anticonvulsant use before and after treatment period. Treatment details including immobilisation technique, number of lesions, dose and fractionation, and volume of lesions were reviewed from department records and the Brainlab iplan system (Brainlab AG, Germany). Prognostic scoring including RTOG RPA, BSBM, SIR, and GPA were calculated (Appendix 1 19 20 21 22 23).

Outcome parameters including local and distant brain control, PFS, and OS were generated using SPSS (Windows version 22.0; IBM Corp, Armonk [NY], US). Univariable analysis with Cox proportional hazards model was performed to generate prognostic factors for local and distinct brain PFS (defined as the time from treatment to documented local progression/distinct brain progression or death) and OS. For each outcome, statistically significant non-modifiable patient and disease factors in univariable analysis together with important treatment factors were included in respective multivariable analysis using Cox proportional hazards model. The enter method was used for variable selection process. Kaplan-Meier survival curve for OS was generated for different prognostic scoring groups and log rank significance was calculated. The study was approved by clinical research ethics committee of the NTEC-CUHK Cluster, Hospital Authority, Hong Kong, with patient informed consent waived.

A total of 68 patients were screened during the study period. Four patients who were treated with fractionated stereotactic radiotherapy and single-fraction SRS in the same treatment were excluded, and thus 64 patients were included. All patients were treated with frameless LINAC-based SRS with ExacTrac system verification, while contouring and dosimetry with the Brainlab iplan system. Dose administered was based on tumour diameter: 22 Gy to lesions of ≤2 cm, 18 Gy to lesions of 2.1-3.0 cm, and 15 Gy to lesions of 3.1-4.0 cm. A 1.5-mm margin was allowed from gross tumour volume to planning target volume. Deviation of dose prescription from departmental protocol was permitted at the individual physician’s discretion.

Among the 64 patients, there were 40 men and 24 women. The median age at the time of treatment was 58 years (range, 22-95 years). The median Eastern Cooperative Oncology Group performance status was 1 (range, 0-3), and Karnofsky Performance Status (KPS) score was 80 (range, 40-100). Primary disease included carcinoma of breast (n=7), lung (n=45), gastrointestinal (n=2), renal cell (n=6), thyroid (n=1), osteosarcoma (n=1), germ cell (n=1), and epithelioid haemangioendothelioma (n=1). Further details of demographics are summarised in Table 1.

A total of 94 lesions were treated with a dose of 12 Gy to 22 Gy according to size (12 Gy, n=6; 15 Gy, n=12; 16 Gy, n=2; 18 Gy, n=48; 20 Gy, n=14; 22 Gy, n=12). The median dose was 18 Gy. The median size of lesion treated was 19 mm (range, 3-43 mm).

The median follow-up time was 11.5 (range, 0.4-56.4) months. One-year actuarial local control rate was 72% (95% confidence interval [CI], 57%-83%) and distant control rate was 71% (95% CI, 56%-82%). The median local PFS was 11.2 (95% CI, 8.4-11.2) months. The median distinct brain PFS was 10.8 (95% CI, 8.4-13.1) months. The median OS was 13.0 (95% CI, 10.6-11.3) months.

Four (6.3%) patients had acute toxicities, mainly brain oedema, and one patient had a seizure for 3 days after treatment. Eight (12.5%) patients had delayed seizure after a median time of 10.5 months. One patient had radionecrosis confirmed pathologically after surgical resection. There were 43 (67.2%) patients who were prescribed steroid before treatment, and eight (12.5%) patients became steroid dependent until their demise. Steroid prescription was not found to affect OS significantly. Nonetheless among the steroid group, becoming steroid dependent was associated with poorer prognosis, with a median OS in the steroid-dependent group of 0.92 months versus 13.6 months in the non–steroid-dependent group (P<0.005, log rank; Appendix 2, Fig a). The worse survival of steroid-dependent patients was independent of volume of brain metastases.

Potential prognostic factors of survival including patient factors such as gender, age, and performance status; and disease factors such as primary cancer, presence of extracranial disease, pre-existing convulsion, number of brain lesions, and size and volume of the largest lesion were examined with reference to decision for SRS treatment by univariable analysis using Cox proportional hazards model. It was found that OS was associated significantly with age (≤70 vs >70 years; P=0.011) and KPS score (>50 vs ≤50; P=0.008). Local PFS was associated significantly with age (≤70 vs >70 years; P=0.043) and KPS score (>50 vs ≤50; P=0.021). Distinct brain PFS was associated significantly with age (≤70 vs >70 years; P=0.02), presence of extracranial disease (presence vs absence; P=0.038), KPS score (>50 vs ≤50; P=0.009), and number of brain lesions (<1-2 vs ≥3; P=0.016). Results of univariable analysis are summarised in Table 2.

Dose relationship for each lesion was analysed separately. Lesions prescribed >18 Gy had statistically significant superior time to progression (radiologically documented local progression) than those given ≤18 Gy, with a 1-year local control rate of 88% vs 60% (Appendix 2, Fig b). Some patients had more than one lesion treated with different doses. Nonetheless after taking into account the highest dose given in the same patient, dose did not affect local PFS or OS significantly (Table 2); dose was not analysed in distinct brain PFS as it should not affect distant brain progression.

With particular reference to the effect of whole-brain radiotherapy (WBRT), it was found that concomitant WBRT (within 3 months of treatment with SRS) did not have a statistically significant impact on OS, local PFS, or distant brain PFS (Table 2).

Multivariable analysis using Cox proportional hazards model and taking patient, disease, and treatment factors into account identified that statistically significant factors associated with distinct brain PFS were KPS score (>50 vs ≤50; P=0.022, hazard ratio [HR]=0.20, 95% CI=0.05-0.80) and number of brain lesions (1-2 vs ≥3; P=0.003, HR=0.31, 95% CI=0.14-0.68) [Table 3].

Of note, a large number of patients in the group had primary lung cancer (n=45), most of which were non–small-cell lung cancer (NSCLC) [n=42]. Among NSCLC patients, a sensitive activating EGFR mutation (exon 19 deletion or exon 21 L858R mutation) was present in 14. Three other patients carried a less common mutation: exon 21 861Q (n=1), exon 18 missense (n=1), and exon 18 179S (n=1). Patients with an exon 19 deletion or exon 21 L858R mutation had superior OS compared with the non-mutational group (P=0.019, HR=0.281, 95% CI=0.097-0.814) but there was no statistically significant difference in local or distant brain control. Among the 14 patients with sensitive activating EGFR mutation, three patients who were diagnosed with brain metastases received WBRT before SRS treatment, and six patients were given SRS together with WBRT. Again, concomitant WBRT was not shown to affect local/distinct brain PFS or OS. For epidermal growth factor receptor (EGFR)–tyrosine kinase inhibitor (TKI) treatment, 12 of 14 patients had lifelong EGFR-TKI treatment, with a median survival of 19.5 months; one with exon 19 deletion and one with exon 18 missense deletion did not have EGFR-TKI treatment. There were seven patients who were prescribed EGFR-TKI before SRS treatment (range of duration, 5.7-21.4 months), and eight patients who were started on or continued on more lines of EGFR-TKI after SRS treatment.

Overall survival was significantly associated with BSBM (P=0.031, log-rank), SIR (P=0.007, log-rank), and GPA (P=0.003, log-rank) [Fig]. A comparison of median survival of the current study with the other original studies is shown in Table 4.19 20 21 22 Of note, DS-GPA was not analysed due to the small number of patients with breast, gastrointestinal, and renal cell primaries. The calculation of GPA and DS-GPA of lung primary was the same.

Brain metastasis has previously been considered an end-of-life event. With the development of new systemic therapies that are effective in both extracranial and intracranial diseases, together with a better understanding from clinical trials of the advantages of SRS, oncologists are more willing to offer SRS to patients with limited brain metastases.

At the other extreme, studies have compared the efficacy of WBRT with supportive care in patients with advanced brain metastases. The latest news from the QUARTZ trial, conducted by the UK Medical Research Council Group, presented at the American Society of Clinical Oncology Meeting in 201524 (full paper awaited) was striking for oncologists. They randomly allocated 538 NSCLC patients with brain metastases that were not amenable to surgery or SRS to either optimal supportive care (OSC) plus WBRT (20 Gy/5 fractions) or OSC alone. There was no significant difference in survival between the OSC+WBRT group and OSC-alone group, with the median survival being 65 and 57 days, respectively. Quality of life was also assessed in this study. The difference between the mean quality-adjusted life-years was -1.9 days only (OSC+WBRT 43.3 vs OSC-alone 41.4 days) and did not meet the initial defined criteria of significance. These data revealed that we are encountering a group of patients with very heterogeneous tumour behaviour and thus personalised treatment is required.

This retrospective study included patients who underwent frameless SRS during January 2010 to July 2015, after commencement of frameless SRS treatment in our centre. Limitations of this study including small number of patients and information bias are inevitable. Nonetheless, the outcomes of patients with brain metastases who underwent frameless SRS in our centre are compatible with those from other large clinical trials that used frame-based systems in terms of control rate, median OS and PFS, and toxicities. Approximately 13% of patients had a complication of steroid dependence that may have been due to treatment or natural disease progression. Steroid dependence was associated with poor survival, independent of volume of tumour. Prolonged use of steroid has been associated with decreased immunity that may underlie superimposed infection. Therefore, tailing down of steroid dose as early as possible in accordance with patient symptoms is strongly recommended.

This study revealed that OS was significantly associated with previously identified prognostic scoring group such as BSBM, SIR, and GPA. Among the three, BSBM and GPA are more convenient to use as only three or four factors are considered respectively, and the information should be easily available in a clinic (including age, KPS, control of primary cancer, presence of extracranial metastases, and number of brain metastases). Data relating to volume of the largest brain lesion included in SIR may not always be available as the reporting radiologist may only report lesion diameter. In terms of patient selection, for patients with GPA of 0-1.0, the median OS was 4.1 months in our study compared with 2.6 months in the original study, and similar to that of patients given WBRT alone. It may be more appropriate to prescribe WBRT alone or best supportive care for this group of patients in lieu of SRS. An important observation from the result of our study is that survival of patients was significantly improved compared with a previous cohort (Table 4). This reflects a significant improvement in systemic treatment over the last decade. Thus, the use of high technology radiation techniques such as SRS is increasingly considered by radiation oncologists to achieve the best outcomes.

Another important aim of this study was to identify prognostic factors of survival in order to avoid futile treatment in those patients who will have a poor outcome despite SRS. Due to the small number of patients in this study, we were not able to identify patients with superior survival among different primaries, similar to DS-GPA. It is of note that a large number of patients in our study had primary lung cancer. In the NSCLC subgroup, patients with an activating EGFR mutation had significantly better survival than those without mutation, and the majority of this group had EGFR-TKI lifelong. Of note, EGFR-TKI has been shown in various studies to have PFS and survival benefit in patients with EGFR-activating mutation.25 26 27 28 29 30 31 32 In a recent retrospective multi-institutional study with more than 300 patients, outcomes of patients with EGFR-activating mutation were analysed following treatment with upfront SRS followed by EGFR-TKI, upfront WBRT followed by EGFR-TKI, and upfront EGFR-TKI.33 Patients in the upfront SRS and upfront WBRT group had significantly superior OS and intracranial PFS compared with those with upfront EGFR-TKI.33 Therefore, in patients with oligo-brain metastases harbouring an EGFR-activating mutation, SRS followed by EGFR-TKI should be considered a standard treatment, and WBRT reserved until there is frank brain disease progression to conserve cognitive function. In addition, SRS combined with efficacious systemic treatment with good brain penetration while omitting WBRT should also be considered in other primaries, although further studies are awaited to validate the benefit.

The beneficial effect of WBRT in addition to SRS is controversial. Recent evidence shows it improves local control but not survival.34 35 Nonetheless, in view of toxicity of somnolence, malaise and cognitive impairment with WBRT, many clinicians may prefer delaying WBRT until there is frank disease progression after SRS. In a recent meta-analysis, the benefit of additional WBRT was not observed in patients who were 50 years old or younger in terms of survival or distant brain control.36 Initial omission of WBRT in this young age-group had no adverse effect on distant brain relapse rate. We were unable to replicate improvement in brain control with WBRT or demonstrate an interaction of age with benefit of concomitant WBRT, possibly due to the small size and retrospective nature of our current study. Number of brain metastases was identified as a significant prognostic factor of brain PFS. Patients with three or more brain metastases had worse PFS than those with one or two brain metastases (5.8 months vs 11.3 months). Again due to the small number of patients, we were unable to demonstrate whether concomitant WBRT could improve brain PFS in patients with three or more brain metastases. Further prospective studies are warranted to verify whether concomitant WBRT should be considered in patients with a higher disease load or age over 50 years.

Frameless SRS for oligo-brain metastases is painless and well tolerated, and should be increasingly considered in patients with good prognostic scores. Its combination with effective systemic treatment has significantly improved survival over the past decade. Nonetheless it is important to individualise treatment for patients with brain metastases according to their inherited prognostic risk factors. High precision treatment with SRS with or without WBRT should be offered to patients with oligo-brain metastases with good prognostic scores and favourable primary histology. For patients with EGFR-activating mutation, SRS followed by EGFR-TKI is a superior choice of treatment. Based on the latest evidence, it may be advisable to give SRS alone and reserve WBRT as salvage for patients with limited brain metastases who are 50 years or younger. Further, WBRT alone can be offered to patients with multiple symptomatic brain metastases and unfavourable prognostic scores. Best supportive care with dexamethasone alone may be considered for patients with very poor performance status.

Frameless SRS is effective and safe for patients with oligo-metastases of brain. Identification of patients with brain metastases who would benefit from SRS is important. Current available prognostic scoring systems provide a good estimation of survival. Frameless SRS should be increasingly considered in patients with favourable prognostic scores.

I would like to thank Dr SF Leung and Dr Kennis Ngar of Department of Clinical Oncology, Prince of Wales Hospital, Hong Kong for their professional opinion and support in this study.

All authors have disclosed no conflicts of interest.

1. Tsao MN, Lloyd NS, Wong RK, et al. Radiotherapeutic management of brain metastases: a systematic review and meta-analysis. Cancer Treat Rev 2005;31:256-73. Crossref

2. Nieder C, Spanne O, Mehta MP, Grosu AL, Geinitz H. Presentation, patterns of care, and survival in patients with brain metastases: what has changed in the last 20 years? Cancer 2011;117:2505-12. Crossref

3. Leksell L. The stereotaxic method and radiosurgery of the brain. Acta Chir Scand 1951;102:316-9.

4. Lo SS. Stereotactic radiosurgery. Available from: http://emedicine.medscape.com/article/1423298-overview#a2. Accessed 19 Mar 2015.

5. Minniti G, Scaringi C, Clarke E, Valeriani M, Osti M, Enrici RM. Frameless linac-based stereotactic radiosurgery (SRS) for brain metastases: analysis of patient repositioning using a mask fixation system and clinical outcomes. Radiat Oncol 2011;6:158. Crossref

6. Ramakrishna N, Rosca F, Friesen S, Tezcanli E, Zygmanszki P, Hacker F. A clinical comparison of patient setup and intra-fraction motion using frame-based radiosurgery versus a frameless image-guided radiosurgery system for intracranial lesions. Radiother Oncol 2010;95:109-15. Crossref

7. ASTRO Policies for Stereotactic radiosurgery (SRS). Available from: https://www.astro.org/uploadedFiles/ Main_Site/Practice_Management/Reimbursement/ASTROSRSModelPolicy.pdf. Accessed 2014.

8. Shaw E, Scott C, Souhami L, et al. Single dose radiosurgical treatment of recurrent previously irradiated primary brain tumors and brain metastases: final report of RTOG protocol 90-05. Int J Radiat Oncol Biol Phys 2000;47:291-8. Crossref

9. Sneed PK, Lamborn KR, Forstner JM, et al. Radiosurgery for brain metastases: is whole brain radiotherapy necessary? Int J Radiat Oncol Biol Phys 1999;43:549-58. Crossref

10. Pirzkall A, Debus J, Lohr F, et al. Radiosurgery alone or in combination with whole-brain radiotherapy for brain metastases. J Clin Oncol 1998;16:3563-9. Crossref

11. Andrews DW, Scott CB, Sperduto PW, et al. Whole brain radiation therapy with or without stereotactic radiosurgery boost for patients with one to three brain metastases: Phase III results of the RTOG 9508 randomised trial. Lancet 2004;363:1665-72. Crossref

12. Cho KH, Hall WA, Gerbi BJ, Higgins PD, Bohen M, Clark HB. Patient selection criteria for the treatment of brain metastases with stereotactic radiosurgery. J Neurooncol 1998;40:73-86. Crossref

13. Varlotto JM, Flickinger JC, Niranjan A, Bhatnagar AK, Kondziolka D, Lunsford LD. Analysis of tumor control and toxicity in patients who have survived at least one year after radiosurgery for brain metastases. Int J Radiat Oncol Biol Phys 2003;57:452-64. Crossref

14. Bhatnagar AK, Flickinger JC, Kondziolka D, Lunsford LD. Stereotactic radiosurgery for four or more intracranial metastases. Int J Radiat Oncol Biol Phys 2006;64:898-903. Crossref

15. Nieder C, Grosu AL, Gaspar LE. Stereotactic radiosurgery (SRS) for brain metastases: a systematic review. Radiat Oncol 2014;9:155. Crossref

16. Pham NL, Reddy PV, Murphy JD, et al. Frameless, real-time, surface imaging-guided radiosurgery: update on clinical outcomes for brain metastases. Transl Cancer Res 2014;3:351-7.

17. Breneman JC, Steinmetz R, Smith A, Lamba M, Warnick RE. Frameless image-guided intracranial stereotactic radiosurgery: clinical outcomes for brain metastases. Int J Radiat Oncol Biol Phys 2009;74:702-6. Crossref

18. Muacevic A, Kufeld M, Wowra B, Kreth FW, Tonn JC. Feasibility, safety, and outcome of frameless image-guided robotic radiosurgery for brain metastases. J Neurooncol 2010;97:267-74. Crossref

19. Gaspar L, Scott C, Rotman M, et al. Recursive partitioning analysis (RPA) of prognostic factors in three Radiation Therapy Oncology Group (RTOG) brain metastases trials. Int J Radiat Oncol Biol Phys 1997;37:745-51. Crossref

20. Lorenzoni J, Devriendt D, Massager N, et al. Radiosurgery for treatment of brain metastases: estimation of patient eligibility using three stratification systems. Int J Radiat Oncol Biol Phys 2004;60:218-24. Crossref

21. Weltman E, Salvajoli JV, Brandt RA, et al. Radiosurgery for brain metastases: a score index for predicting prognosis. Int J Radiat Oncol Biol Phys 2000;46:1155-61. Crossref

22. Sperduto PW, Berkey B, Gaspar LE, Mehta M, Curran W. A new prognostic index and comparison to three other indices for patients with brain metastases: an analysis of 1,960 patients in the RTOG database. Int J Radiat Oncol Biol Phys 2008;70:510-4. Crossref

23. Sperduto PW, Kased N, Roberge D, et al. Summary report on the graded prognostic assessment: an accurate and facile diagnosis-specific tool to estimate survival for patients with brain metastases. J Clin Oncol 2012;30:419-25. Crossref

24. Mulvenna PM, Nankivell MG, Barton R, et al. Whole brain radiotherapy for brain metastases from non-small lung cancer: Quality of life (QoL) and overall survival (OS) results from the UK Medical Research Council QUARTZ randomised clinical trial (ISRCTN 3826061). J Clin Oncol 2015;33(Suppl);abstract8005.

25. Mok TS, Wu YL, Thongprasert S, et al. Gefitinib or carboplatin-paclitaxel in pulmonary adenocarcinoma. N Engl J Med 2009;361:947-57. Crossref

26. Han JY, Park K, Kim SW, et al. First-SIGNAL: first-line single-agent iressa versus gemcitabine and cisplatin trial in never-smokers with adenocarcinoma of the lung. J Clin Oncol 2012;30:1122-8. Crossref

27. Mitsudomi T, Morita S, Yatabe Y, et al. Gefitinib versus cisplatin plus docetaxel in patients with non-small-cell lung cancer harbouring mutations of the epidermal growth factor receptor (WJTOG3405): an open label, randomised phase 3 trial. Lancet Oncol 2010;11:121-8. Crossref

28. Maemondo M, Inoue A, Kobayashi K, et al. Gefitinib or chemotherapy for non-small-cell lung cancer with mutated EGFR. N Engl J Med 2010;362:2380-8. Crossref

29. Jänne PA, Wang X, Socinski MA, et al. Randomized phase II trial of erlotinib alone or with carboplatin and paclitaxel in patients who were never or light former smokers with advanced lung adenocarcinoma: CALGB 30406 trial. J Clin Oncol 2012;30:2063-9. Crossref

30. Rosell R, Carcereny E, Gervais R, et al. Erlotinib versus standard chemotherapy as first-line treatment for European patients with advanced EGFR mutation–positive non-small-cell lung cancer (EURTAC): a multicentre, open-label, randomised phase 3 trial. Lancet Oncol 2012;13:239-46. Crossref

31. Zhou C, Wu YL, Chen G, et al. Erlotinib versus chemotherapy as first-line treatment for patients with advanced EGFR mutation–positive non-small-cell lung cancer (OPTIMAL, CTONG-0802): a multicentre, open-label, randomised, phase 3 study. Lancet Oncol 2011;12:735-42. Crossref

32. Yang JC, Wu YL, Schuler M, et al. Afatinib versus cisplatin-based chemotherapy for EGFR mutation–positive lung adenocarcinoma (LUX-Lung 3 and LUX-Lung 6): analysis of overall survival data from two randomised, phase 3 trials. Lancet Oncol 2015;16:141-51. Crossref

33. Magnuson WJ, Lester-Coll NH, Wu AJ, et al. Management of brain metastases in tyrosine kinase inhibitor-naïve epidermal growth factor receptor-mutant non-small-cell lung cancer: a retrospective multi-institutional analysis. J Clin Oncol 2017;35:1070-7. Crossref

34. Mehta MP, Tsao MN, Whelan TJ, et al. The American Society for Therapeutic Radiology and Oncology (ASTRO) evidence-based review of the role of radiosurgery for brain metastases. Int J Radiat Oncol Biol Phys 2005;63:37-46. Crossref

35. Patil CG, Pricola K, Sarmiento JM, Garg SK, Bryant A, Black KL. Whole brain radiation therapy (WBRT) alone versus WBRT and radiosurgery for the treatment of brain metastases. Cochrane Database Syst Rev 2012;(9):CD006121. Crossref

36. Sahgal A, Aoyama H, Kocher M, et al. Phase 3 trials of stereotactic radiosurgery with or without whole-brain radiation therapy for 1 to 4 brain metastases: individual patient data meta-analysis. Int J Radiat Oncol Biol Phys 2015;91:710-7. Crossref

The World Health Organization (WHO) considers iodine deficiency the single most important preventable cause of brain damage worldwide. Since 1993, WHO has recommended universal salt iodisation to eliminate iodine deficiency disorder.1 There is a common belief that the Hong Kong population residents in coastal regions should have sufficient iodine intake. Twenty years ago, however, 50% of children and adults were found iodine-deficient according to the WHO standards.2 At the same time, studies also reported that one third of local pregnant women were iodine-deficient based on their urinary iodine concentration (UIC), from the first through third trimester.3 4 Pregnant and lactating women are among the most vulnerable groups in the population as iodine plays an important role in early neuronal migration and maturation in the developing fetus and infants.

A local expert panel group was established in 2003 to encourage the monitoring of iodine status and rectify the problem.5 The panel concluded that dietary iodine intake of our population was borderline sufficient, and was also inadequate to meet any extra requirement at the time of thyroid stress, ie pregnancy, neonatal period, and in the first few years of life.5 In a local population survey conducted by the Centre for Food Safety during 2005 to 2007, the median iodine dietary intake was as low as 44 µg/day among 5000 adults, while 60% had an intake of <50 µg/day (the threshold for normal thyroid functioning).6 Only 5% had an iodine intake within the safe range. This report, however, also admitted a lack of clinical and biochemical data (ie UIC) that precluded a conclusion about iodine deficiency. Since this report in 2011, all mothers are enquired about their dietary intake of iodine and provided with iodine-containing multivitamin supplements when they book in at a Maternity and Child Health Centre in Hong Kong.

Nevertheless, the problem remains unresolved as two recent cohort studies in the UK and Australia consistently reported that low maternal first-trimester UIC was associated with poorer childhood cognitive development at the age of 8 to 9 years.7 8 The investigators of the Avon Longitudinal Study of Parents and Children in the UK also reported a dose-effect association; child neurocognitive scores were lower when maternal urinary iodine-to-creatinine ratio (UICr) was <50 µg/g.7

As no salt iodisation programme has been implemented since the first study was carried out, and with the changes in social circumstances and dietary trends over the last two decades, it was deemed necessary and important to reassess the iodine status of pregnant and lactating women in Hong Kong to guide future policy. The objective of this paper was to report the iodine status of women during early gestation at an obstetric unit in Hong Kong.

Between July 2014 and November 2015, healthy pregnant women were enrolled at the antenatal clinic of Prince of Wales Hospital into a study of gestational age-specific thyroid function test reference intervals. Women without a history of thyroid dysfunction, hyperemesis gravidarum, autoimmune disease, or any other major medical conditions were eligible. Consecutive cases were recruited at their first attendance for antenatal booking. We estimated that 250 blood samples were required for each block of gestational age range (<6, 6-10, >10-14, >14-18, >18-24, >24-32, >32-38, and >38 weeks) to generate a nomogram; 600 subjects were required to provide 2000 samples, each had an average of four blood samples across the gestation, with an estimated dropout rate of 15%. All subjects underwent early ultrasound scan for dating with gestational age adjusted according to the ultrasound date if it differed to that calculated from the last menstrual period. The study was approved by the Chinese University of Hong Kong Clinical Research Ethics Committee (CRE-2013.500), and written informed consent was obtained from all women.

As part of the study design, UIC was measured at the time of recruitment to determine the participant’s iodine status. Women were asked to collect a morning urine sample into an acid-washed trace element urine bottle within 1 week of recruitment for the assay. In this study, UIC was measured by an inductively coupled plasma mass spectrometer (ICP-MS 7700; Agilent Technologies, US). The intra- and inter-assay coefficients of variation for UIC were 3.3% and 7.4% at 11.9 µg/L, and 2.7% and 4.9 % at 49.6 µg/L, respectively. Urinary creatinine concentration was also measured by IDMS-traceable Jaffe kinetic reaction (cobas 8000; Roche Diagnostics Inc, US) to calculate the UICr.

All participants were interviewed by a research nurse who asked about the duration, frequency, and brand of any multivitamin supplements taken to quantify daily iodine supplementation. Due to the limited funding, dietary intake of iodine-containing foods was only assessed in a conveniently sampled subgroup of the population by another research assistant using a food frequency questionnaire (FFQ; Appendix). Education level and occupation were recorded for this subgroup. The FFQ was based on the literature and previously validated FFQs that have been used in the local population.6 9 10 Eleven main food groups that contribute to iodine intake in the local population were incorporated into the FFQ (Table 1). Portions were reported based on standard reference sizes, eg cups, grams, centilitres. To facilitate understanding and recall, food photo albums and eating utensils of standard portions were presented when completing the FFQ. Use of iodised salt was also documented. Daily dietary intake of iodine was calculated using Food Processor Nutrition Analysis and Fitness software, version 8.0 (ESHA Research, Salem, US). The iodine content of local foods was also programmed into the software, extracted from food composition tables from Hong Kong SAR and Mainland China.11 We used the WHO definition to categorise insufficient, adequate, above requirements, and excessive by UIC of <150, 150-249, 250-499, and ≥500 µg/L, respectively12; UICr of <150 µg/g was also used to define insufficiency according to Bath et al.7 An average daily iodine intake of 250 µg/day was considered adequate for pregnancy.1 Data are expressed as mean ± standard deviation (SD), median and interquartile range (IQR), or counts with proportion. Between-group differences were compared using the Student’s t test and χ²/Fisher’s exact tests, as appropriate. Multivariate logistic regression analyses were used to obtain adjusted odds ratios with 95% confidence intervals, with the forced entry of covariates, namely maternal age, parity, education level, occupation, body mass index, and gestational age at recruitment, to assess factors associated with low UIC (<150 µg/L), low UICr (<150 µg/g) and iodine-containing supplementation. Multivariate linear regression analyses were used to assess the association with UIC and UICr. Statistical analysis was performed using SPSS (Windows version 20.0; IBM Corp, Armonk [NY], US). A P value of <0.05 was used to indicate significance for two-tailed statistical test results. The gestational age-specific thyroid function test reference intervals are not included in this article but will be published elsewhere.

A total of 600 healthy pregnant women were enrolled during the study period at a median gestational age of 7.0 (IQR, 5.9-8.6) weeks. The mean (± SD) age and body mass index at recruitment were 31.3 ± 3.9 years and 21.9 ± 3.1 kg/m², respectively. Of the subjects, 382 (63.7%) women were nulliparous and two (0.3%) were twin pregnancies. The median (IQR) UIC and UICr were 100 (58-165) µg/L and 98 (67-150) µg/g, respectively (Table 2). According to the WHO definition, 429 (71.5%) participants were regarded as iodine deficient (Fig a). Moreover, 450 (75.0%) participants had UICr of <150 µg/g (Fig b).7

Daily dietary intake of iodine was assessed in 146 participants. They were slightly younger in age and had a higher rate of nulliparity; their UIC and UICr were no different to all participants (Table 2). The median (IQR) daily dietary intake of iodine was 69.5 (47.3-152.4) µg. The greatest source of iodine among the studied sample was from seaweeds that were consumed by 76% of women (Table 1). The next most common source of iodine was non-alcoholic beverages such as soy milk, soup, soft drinks, and water (21.5% of the daily iodine intake). Of all the 146 women surveyed, only four (2.7%) regularly used iodised salt in their diet. Several food sources were commonly eaten by all women including fish, cereal and grain products, vegetables and legumes, and condiments. Nonetheless these combined food sources only contributed to 15.5% of daily dietary iodine intake. There were 44 (30.1%), 92 (63.0%) and 122 (83.6%) participants who had a daily iodine intake below 50, 100, and 250 µg respectively.

Nutritional supplementation was reported by 414 (69.0%) of all participants at the time of enrolment but only 163 (39.4%) of these supplements contained iodine; the daily iodine supplement was between 140 and 290 µg if taken according to the prescription. The subgroup of participants who had taken iodine-containing supplements for 2 weeks or more had an adequate iodine status both by their median UIC and UICr (Table 3). In 122 subjects who completed the FFQ and had urinary iodine level measured within 1 week of each other, total daily iodine intake (iodine supplementation included) had a weak correlation with UICr (r=0.20, P=0.03). In multivariate regression analyses, no covariates except gestational age at recruitment were associated with low UIC or UICr, or iodine supplementation; women recruited at a later gestational age had significantly higher UIC and were more likely to have started taking an iodine-containing supplement (Table 4).

The findings from this study suggest that local pregnant women might still be iodine-deficient when the WHO definition is applied. In the subgroup of participants who completed the FFQ, more than 80% did not have an adequate daily iodine intake as recommended by the WHO and other authorities.13 14 This is in keeping with the survey conducted by the Centre for Food Safety 10 years ago. A recent local study of 95 lactating mothers showed that the mean daily dietary intake of iodine was 62.6 µg with only 2% having a sufficient iodine intake.15 The result also showed that only 48% of breast milk samples from 39 women with an infant younger than 6 months had an adequate level of iodine (85 µg per day) as recommended by the Chinese Dietary Reference Intake.15 The revised guideline from the Department of Health in 2016 suggested that mothers consider taking iodine-containing prenatal supplements during pregnancy and breastfeeding in view of the difficulties in obtaining sufficient iodine from food alone.16 The policy appears to be in agreement with our findings that mothers taking iodine-containing supplements were iodine sufficient according to their median UIC. Nonetheless universal iodine supplementation remains controversial. The American Thyroid Association, Endocrine Society, and New Zealand Ministry of Health recommend that women who are planning a pregnancy or who are currently pregnant or lactating should receive 150 µg per day of iodine supplementation in the form of potassium iodide-containing supplements.17 18 On the contrary, Cochrane systematic review did not support routine iodine supplementation in women before, during or after pregnancy, while WHO does not recommend iodine supplementation in regions where the median UIC indicates iodine sufficiency, or where a salt iodisation programme is in place.1 19 All eight iodine-containing supplements that were reportedly taken by participants in this study also contained iron, and frequently resulted in varying degrees of constipation. Most women did not start taking supplements prior to their first antenatal visit. Even in those who did, 60% of the supplements contained no iodine, similar to that reported in the US.20 21 Our results also show that mothers were more likely to have commenced iodine supplementation and have an iodine adequate status at a later gestation; this suggests mothers might acquire the knowledge as pregnancy progresses.

Although the UIC is designated for assessment of a population, not individual iodine status, our results suggest that three quarters of the participants may have been iodine insufficient during early pregnancy according to their UIC and UICr. Similarly, iodine deficiency in pregnancy has re-emerged in several developed countries.22 Subjects enrolled into the Avon study in the UK were mildly-to-moderately iodine deficient with a median UIC of 91 µg/L.7 A more recent study among schoolgirls in the UK also reported a similarly low UIC suggesting that mild-to-moderate iodine deficiency remains a problem.23 Such deficiency was probably caused by a poor availability of iodised salt, few UK recommendations for increased iodine intake in pregnancy, and insufficient use of iodine-containing prenatal supplements. The International Council for Control of Iodine Deficiency Disorders global network now places the UK on the list of mildly deficient nations.24 The Department of Health in the UK has also added iodine to its National Diet and Nutrition Survey that checks the nutrient intake of adults and children in the UK.25 In Australia, mandatory iodine fortification of salt used in bread was introduced in 2009 in order to tackle mild iodine deficiency in the population; iodine status of women has improved since then but only those taking iodine-containing supplements have UIC indicative of sufficiency.26 According to data from the National Health and Nutrition Examination Survey 2005-2010, more than 55% of pregnant women had UIC that suggested inadequate iodine intake.27 Mainland China has led the way in sustaining the elimination of iodine deficiency through iodised salt since the early 1990s.28 Nonetheless dietary iodine intake remains insufficient among pregnant women in Shanghai, Zhejiang, and other coastal cities where iodine is considered sufficient for the general population.29 30 A national survey from the Mainland reported that the median UIC levels of pregnant and lactating women were 123-224 µg/L and 109-224 µg/L, respectively; median UIC was higher among those in inland cities, because of a higher iodine level in the salt and greater household coverage.31 In contrast, the progress in tackling iodine insufficiency in Hong Kong has been far from satisfactory.

Due to the limitations in study design, this study can only provide a snapshot overview of the iodine status of local pregnant women during early pregnancy from a single obstetric unit. Given that the main objective of the study was to ensure our study group had sufficient iodine intake in order to establish a thyroid function test reference range for the local population, our results cannot draw any conclusions about iodine status during the second and third trimesters. Moreover, dietary iodine intake could only be assessed in a subgroup of the population due to limited funding. The WHO has considered iodine deficiency the single most important preventable cause of brain damage worldwide. Recent study also highlighted the impact of iodine deficiency in the first 1000 days of life especially among breast-fed infants.32 As the Hong Kong SAR Government is actively promoting breastfeeding, it is also important to ensure adequate iodine in lactating mothers and breast-fed infants. It is time to systematically revisit the iodine status of our local women at pre-conception, and during pregnancy and lactation.

Results of our study are in line with those from the survey by the Centre for Food Safety, suggesting that our local pregnant women are borderline iodine-deficient and have an inadequate dietary iodine intake during early pregnancy. There is a need to educate the public and to advise women of childbearing age to maintain sufficient dietary iodine before contemplating pregnancy. A policy of salt iodisation and regular monitoring of iodine status with UIC in our population should be considered.

This study was supported by the Hospital Authority of Hong Kong on a project to derive a gestational age-specific thyroid function reference interval for the local pregnant population. Ms Sharon Lai-kwai Chan, research nurse, recruited all subjects, and collected and organised the clinical data. Ms Macy Kwan conducted face-to-face interviews for the dietary questionnaire for the subgroup of subjects.

All authors have disclosed no conflicts of interest.

1. World Health Organization; United Nations Children's Fund; International Council for the Control of Iodine Deficiency Disorders. Assessment of iodine deficiency disorders and monitoring their elimination: A guide for programme managers. 3rd ed. Available from: http://apps.who.int/iris/bitstream/10665/43781/1/9789241595827_eng.pdf. Accessed 22 Mar 2017.

2. Kung AW, Chan LW, Low LC, Robinson JD. Existence of iodine deficiency in Hong Kong—a coastal city in southern China. Eur J Clin Nutr 1996;50:569-72.

3. Kung AW, Lao TT, Low LC, Pang RW, Robinson JD. Iodine insufficiency and neonatal hyperthyrotropinaemia in Hong Kong. Clin Endocrinol (Oxf) 1997;46:315-9. Crossref

4. Kung AW, Lao TT, Chau MT, Tam SC, Low LC. Goitrogenesis during pregnancy and neonatal hypothyroxinaemia in a borderline iodine sufficient area. Clin Endocrinol (Oxf) 2000;53:725-31. Crossref

5. But B, Chan CW, Chan F, et al. Consensus statement on iodine deficiency disorders in Hong Kong. Hong Kong Med J 2003;9:446-53.

6. Centre for Food Safety, Hong Kong SAR Government. Risk Assessment Studies Report No. 45: Dietary iodine intake in Hong Kong adults. Jul 2011. Available from: http://www.cfs.gov.hk/english/programme/programme_rafs/programme_rafs_n_01_12_Dietary_Iodine_Intake_HK.html. Accessed 20 Apr 2015.

7. Bath SC, Steer CD, Golding J, Emmett P, Rayman MP. Effect of inadequate iodine status in UK pregnant women on cognitive outcomes in their children: results from the Avon Longitudinal Study of Parents and Children (ALSPAC). Lancet 2013;382:331-7. Crossref

8. Hynes KL, Otahal P, Hay I, Burgess JR. Mild iodine deficiency during pregnancy is associated with reduced educational outcomes in the offspring: 9-year follow-up of the gestational iodine cohort. J Clin Endocrinol Metab 2013;98:1954-62. Crossref

9. Combet E, Lean ME. Validation of a short food frequency questionnaire specific for iodine in U.K. females of childbearing age. J Hum Nutr Diet 2014;27:599-605. Crossref

10. Condo D, Makrides M, Skeaff S, Zhou SJ. Development and validation of an iodine-specific FFQ to estimate iodine intake in Australian pregnant women. Br J Nutr 2015;113:944-52. Crossref

11. 楊月欣. 食物營養成分速查 [in Chinese]. 人民日報出版社; 2006.

12. World Health Organization. Urinary iodine concentrations for determining iodine status in populations. 2013. Available from: http://www.who.int/vmnis/indicators/urinaryiodine/en/. Accessed 22 Apr 2015.

13. Institute of Medicine (US) Panel on Micronutrients. Dietary reference intakes for vitamin A, vitamin K, arsenic, boron, chromium, copper, iodine, iron, manganese, molybdenum, nickel, silicon, vanadium, and zinc. Washington, DC: National Academies Press; 2001. Available from: https://www.ncbi.nlm.nih.gov/books/NBK222310/. Accessed 22 Mar 2017.

14. American Thyroid Association. Iodine deficiency. 4 Jun 2012. Available from: http://www.thyroid.org/iodine-deficiency/. Accessed 25 Mar 2015.

15. PolyU discovers inadequate calcium, iron and iodine intakes of Hong Kong lactating women. 26 Jul 2016. Available from: https://www.polyu.edu.hk/web/en/media/media_releases/index_id_6237.html. Accessed 25 Jun 2017.

16. Family Health Service, Department of Health, Hong Kong SAR Government. Healthy eating during pregnancy and breastfeeding. Revised Nov 2016. Available from: http://www.fhs.gov.hk/english/health_info/woman/20036.html. Accessed 21 Jun 2017.

17. Alexander EK, Pearce EN, Brent GA, et al. 2017 Guidelines of the American Thyroid Association for the Diagnosis and Management of Thyroid Disease During Pregnancy and the Postpartum. Thyroid 2017;27:315-89. Crossref

18. Ministry for Primary Industries; Ministry of Health, New Zealand. Mandatory iodine fortification in New Zealand: Supplement to the Australian Institute of Health and Welfare 2016 report—Monitoring the health impacts of mandatory folic acid and iodine fortification. MPI Technical—Paper No: 2016/32. Available from: https://www.mpi.govt.nz/dmsdocument/12786-mandatory-iodine-fortification-in-new-zealand-supplement-to-the-australian-institute-of-health-and-welfare-2016-report-monitoring-the-health-impacts-of-mandatory-folic-acid-and-iodine-fortification. Accessed 22 Mar 2017.

19. Harding KB, Peña-Rosas JP, Webster AC, et al. Iodine supplementation for women during the preconception, pregnancy and postpartum period. Cochrane Database Syst Rev 2017;(3):CD011761. Crossref

20. Gregory CO, Serdula MK, Sullivan KM. Use of supplements with and without iodine in women of childbearing age in the United States. Thyroid 2009;19:1019-20. Crossref

21. Leung AM, Pearce EN, Braverman LE. Iodine content of prenatal multivitamins in the United States. N Engl J Med 2009;360:939-40. Crossref

22. Li M, Eastman CJ. The changing epidemiology of iodine deficiency. Nat Rev Endocrinol 2012;8:434-40. Crossref

23. Vanderpump MP, Lazarus JH, Smyth PP, et al. Iodine status of UK schoolgirls: a cross-sectional survey. Lancet 2011;377:2007-12. Crossref

24. Iodine global network. Global Scorecard 2014: Number of iodine deficient countries more than halved in past decade. IDD Newsletter 2015;Feb:5-7. Available from: http://www.ign.org/cm_data/IDD_feb15_mail.pdf. Accessed 25 Mar 2015.

25. Whitton C, Nicholson SK, Roberts C, et al. National Diet and Nutrition Survey: UK food consumption and nutrient intakes from the first year of the rolling programme and comparisons with previous surveys. Br J Nutr 2011;106:1899-914. Crossref

26. Charlton KE, Yeatman H, Brock E, et al. Improvement in iodine status of pregnant Australian women 3 years after introduction of a mandatory iodine fortification programme. Prev Med 2013;57:26-30. Crossref

27. Caldwell KL, Pan Y, Mortensen ME, Makhmudov A, Merrill L, Moye J. Iodine status in pregnant women in the National Children's Study and in U.S. women (15-44 years), National Health and Nutrition Examination Survey 2005-2010. Thyroid 2013;23:927-37. Crossref

28. Sun D, Codling K, Chang S, et al. Eliminating iodine deficiency in China: achievements, challenges and global implications. Nutrients 2017;9.pii:E361. Crossref

29. Zou S, Wu F, Guo C, et al. Iodine nutrition and the prevalence of thyroid disease after salt iodization: a cross-sectional survey in Shanghai, a coastal area in China. PLoS One 2012;7:e40718. Crossref

30. Wang Z, Zhu W, Mo Z, et al. An increase in consuming adequately iodized salt may not be enough to rectify iodine deficiency in pregnancy in an iodine-sufficient area of China. Int J Environ Res Public Health 2017;14:piiE206.

31. National Food Safety and Risk Assessment Expert Committee. Salt iodization and risk assessment of iodine status in Chinese population [in Chinese]. Technical report No. 2010-002. May 2010. Available from: http://www.nhfpc.gov.cn/cmsresources/mohwsjdj/cmsrsdocument/doc9250.pdf. Accessed 21 Jun 2017.

32. Stinca S, Andersson M, Herter-Aeberli I, et al. Moderate-to-severe iodine deficiency in the “first 1000 days” causes more thyroid hypofunction in infants than in pregnant or lactating women. J Nutr 2017;147:589-95. Crossref

The diagnosis of Cushing’s syndrome, especially when hypersecretion is mild, is plagued by uncertainties. Most clinical features—such as diabetes mellitus, hypertension, obesity, hyperlipidaemia, osteoporosis, or depression—are non-specific and highly prevalent in the general population. More specific features such as myopathy or easy bruising may be absent even in subjects with florid biochemical hypercortisolism. In addition, no single test can diagnose Cushing’s syndrome with 100% sensitivity and specificity.1 It is a common phenomenon that tests for hypercortisolism—for examples, 24-hour urinary free cortisol (UFC), late-night salivary cortisol (SalF_LN) or serum cortisol level after 1-mg overnight dexamethasone suppression test (SerF_Dex)—often produce discordant results. Each test has its own caveats, affected by the level of binding proteins, completeness of urine collection, and absorption and metabolism of dexamethasone. In recent years, increased awareness of the metabolic and cardiovascular consequences of Cushing’s syndrome2 has led to a pressing need to identify tests that are easy to perform and can provide useful information at a low cost.

Collection of saliva samples is non-invasive and stress-free.3 4 Salivary cortisol is not affected by the levels of binding proteins so it provides a reliable indication of the biologically active free serum cortisol level.5 Significant advances have been made with the use of salivary cortisol in the investigation of hypercortisolism.3 6 7 8 9 The availability of liquid chromatography–tandem mass spectrometry (LC-MS/MS) also enables the measurement of other glucocorticoid analytes. Among these and of particular interest is cortisone that is present in the saliva at a higher concentration—the salivary cortisone–to–cortisol ratio is up to 6-8:110 11 12—due to conversion of cortisol to cortisone by the 11 beta-hydroxysteroid dehydrogenase 2 (11β-HSD2) enzyme in the salivary glands.13 This higher concentration makes it more detectable than salivary cortisol.10 Salivary cortisone has been shown by some investigators to have a better and more linear relationship with serum total cortisol14 and serum free cortisol10 than salivary cortisol.

In this study, we reviewed the late-night salivary cortisone (SalE_LN) and post-overnight dexamethasone suppression salivary cortisone (SalE_Dex) values of subjects being investigated for hypercortisolism in our centre, in an attempt to define cut-off values with reasonable sensitivity and specificity.

All subjects referred to the Endocrine Unit of Queen Elizabeth Hospital in Hong Kong for suspected endogenous hypercortisolism were evaluated according to a pre-specified protocol. Written informed consent was obtained from all subjects. The results of subjects investigated during May 2013 (when salivary measurement by LC-MS/MS became available) to September 2016 were reviewed. This study was approved by the Hospital Ethics Committee. No patient was receiving medical treatment for Cushing’s syndrome at the time of assessment. Subjects who were taking medication (such as rifampicin, phenytoin, phenobarbital, and alcohol) that might interfere with dexamethasone metabolism, or were night or shift workers, were excluded.

Detailed oral and written instructions were given to all subjects by an endocrine specialist nurse. For the collection of saliva sample, subjects were instructed to refrain from smoking, brushing teeth, and eating or drinking anything but water for at least 30 minutes before collection. Saliva samples were collected using a cotton swab in Salivette tubes (Sarstedt, Nümbrecht, Germany). Salivettes were kept at 4°C in a home refrigerator and sent to the laboratory within 24 hours.

According to the pre-specified protocol, on day 1, a 24-hour urine sample was collected for UFC measurement. On day 2, a 0900h saliva sample was collected at the Endocrine Centre, under nurse supervision. The patient was then instructed to collect a late-night (between 2300h and 2400h) saliva sample that evening, after which he/she was to take dexamethasone 1 mg orally. On day 3, subjects returned to the Endocrine Centre for a simultaneous blood and saliva sample at 0900h. The blood sample was sent for serum cortisol assay. Saliva samples were assayed for both cortisol and cortisone.

Serum cortisol was measured by competitive chemiluminescent microparticle immunoassay using the Abbott ARCHITECT i2000SR system (Abbott Diagnostics, Illinois, US). The coefficient of variation of the assay for serum cortisol was 4.0%-6.2% at low levels and 3.3%-4.3% at high levels. Salivary cortisol and salivary cortisone were measured by LC-MS/MS using the Waters Xevo TQ MS system (Waters Corporation, Milford [MA], US). Cortisol and cortisone were extracted from saliva using the organic solvent methyl tert-butyl ether after addition of a deuterium-labelled internal standard mixture of cortisol-d4 and cortisone-d8 (CDN isotopes). The organic supernatant was dried under nitrogen at a temperature below 45°C and dissolved in the initial mobile phase for LC-MS/MS analysis. The steroid analytes were separated from the matrix background in a reversed-phase chromatography that employed a sub-2 μm analytical column (ACQUITY UPLC HSS T3 Column, 2.1 x 100 mm, 1.8 mm; Waters Corporation, Milford [MA], US) and a 6-minute elution method consisting of a gradient mixture of 0.1% glacial acetic acid, 0.2 mM ammonium acetate in water and methanol. Negative electrospray mode was used for analyte ionisation. The charged acetate adducts were monitored by multiple reaction monitoring mode with two stable mass transitions for cortisol (421>331; 421>297) and cortisone (419>329; 419>301) and one multiple reaction monitoring for each of the corresponding deuterated internal standards (cortisol-d4: 425>335; cortisone-d8: 427>337). Quantitative measurement was derived using the linear least squares regression method with origin excluded and 1/x weighting for better accuracy at a low concentration level. The coefficient of variation of the assay for salivary cortisol and cortisone was 5%-7% and 7%-10%, respectively across the analyte reporting ranges up to 250 nmol/L. The lower limit of detection was 0.5 nmol/L for both salivary cortisol and cortisone. Urinary cortisol was also measured by LC-MS/MS. Adrenocorticotropic hormone (ACTH) was measured by Immulite 2000 XPi (Siemens Healthcare GmbH, Erlangen, Germany) chemiluminescent immunometric assay. The upper reference limit of ACTH was 10.2 pmol/L.

Subjects were classified as having hypercortisolism if either the biochemical or the histological criterion was fulfilled. The biochemical criterion was defined as having an abnormal value in at least two of the following three tests: (1) SerF_Dex >138 nmol/L, or >50 nmol/L in the context of adrenal incidentaloma15; (2) UFC >157 nmol/day; and (3) SalF_LN ≥3 nmol/L. The categorisation of hypercortisolism was made without knowledge (ie blinded) of the three outcome parameters being evaluated for diagnostic accuracy, namely SalE_LN, post-overnight dexamethasone suppression salivary cortisol (SalF_Dex), and SalE_Dex. The reference range for UFC in our laboratory, established locally from the 2.5th to 97.5th percentile of 112 healthy adults, was 22-157 nmol/day. The reference level for SalF_LN in our laboratory, derived from the 97.5th percentile of 61 normal individuals, was <3 nmol/L. The histological criterion was defined as histological proof and postoperative improvement in biochemical and clinical parameters. Subjects not fulfilling either of these criteria were classified as having eucortisolism.

For calculation purposes, results below the detection limit of the assay were set to the lowest detection value. Continuous data were expressed as mean ± standard deviation if parametric, and median (range) if non-parametric. Chi squared test was used to detect any difference between categorical data. Unpaired t test was used to compare continuous variables, and Mann-Whitney test was used for non-parametric data. A P value of <0.05 was considered statistically significant. Correlation between serum and salivary values were performed using Pearson correlation coefficients.

For estimation of the optimal diagnostic cut-off value, receiver operating characteristic (ROC) curves were drawn using data from the subjects classified as hypercortisolism or eucortisolism. The optimal cut-off was chosen where the Youden’s index (sensitivity + specificity–1) was maximal. The test performance characteristics were calculated to assess their utility. The quality of diagnostic tests was expressed as the area under ROC curve (AUC). For sample size requirement estimation, for an estimated AUC of 0.8, a minimum of nine positive cases was required.16 Statistical analysis was performed using the Statistical Package for Social Science (Windows version 20.0; IBM Corp, Armonk [NY], US).

A total of 115 subjects were referred to our Endocrine Clinic for assessment of hypercortisolism during the study period. Of them, 14 subjects with a history of transsphenoidal surgery or adrenalectomy who had been referred for postoperative assessment of endocrine function were excluded. One patient with ongoing investigations and pending re-evaluation and another with end-stage renal failure were also excluded. No patient was taking exogenous steroids. All subjects had normal renal and liver function tests.

A total of 115 sets of biochemical investigations were performed in 99 subjects (40 males, 59 females; mean age, 55.3 ± 14.3 years; range, 19-81 years). The primary indications for testing were adrenal incidentaloma in 52 subjects, suspected secondary hypertension or diabetes mellitus in 25, Cushingoid features in 21, and pituitary mass in one. Eleven subjects had two or more sets of investigations performed (two patients had 4 sets, one patient had 3 sets, and eight patients had 2 sets). For these subjects, only the data set with the highest SerF_Dex was chosen for analysis. In two subjects, the volume of the late-night salivary sample was inadequate for measurement.

In this study, 21 subjects were found to have hypercortisolism according to the above criteria—20 subjects met the biochemical criterion; eight subjects met the histological criterion, including one whose set of tests did not fulfil the biochemical criterion (SerF_Dex 135 nmol/L; normal UFC and SalF_LN, ACTH 1.1 pmol/L) and who underwent surgery because of an adrenal adenoma that enlarged from 1.6 cm to 2.5 cm over 2 years, postoperative spot cortisol was <28 nmol/L and hydrocortisone replacement was required for 6 months before axis recovery. Among the 21 subjects who had hypercortisolism, 7 had adrenal Cushing’s, 4 pituitary Cushing’s, 3 ectopic ACTH syndrome, 1 adrenocortical carcinoma, and 6 subclinical Cushing’s. Of these subjects, 14 (67%) had elevated UFC, 18 (86%) had non-suppressed SerF_Dex, and 17 (81%) had elevated SalF_LN. Among the eight subjects with histological proof (6 adrenalectomies, 1 transsphenoidal surgery, 1 enucleation of pancreatic neuroendocrine tumour), all had clinical and biochemical improvement after operation. Eucortisolism was diagnosed in 78 subjects according to the aforementioned criteria.

The baseline characteristics of the subjects are shown in Table 1. There were no statistically significant differences between the hypercortisolism and the eucortisolism groups with respect to age, gender, and prevalence of obesity, diabetes mellitus, or hypertension. There was a statistically significantly higher prevalence of Cushingoid features and of proximal myopathy in the hypercortisolism group.

The biochemical test results are shown in Table 2. The hypercortisolism group had statistically significantly higher levels of SerF_Dex, UFC, SalF_LN, SalE_LN, SalF_Dex, and SalE_Dex. No SalF_LN sample in the hypercortisolism group and 17 (21.8%) SalF_LN samples in the eucortisolism group had levels below the detection limit of 0.5 nmol/L. No SalE_LN sample in the hypercortisolism group had levels below the detection limit of 0.5 nmol/L. Four (19.0%) SalF_Dex samples in the hypercortisolism group and 68 (87.2%) SalF_Dex samples in the eucortisolism group had a level below the detection limit of 0.5 nmol/L. One (1.3%) SalE_Dex sample in the eucortisolism group had a level below the detection limit of 0.5 nmol/L.

The ROC analysis (Fig 1) and Table 3 reveal that the optimal cut-off for SalE_LN was 13.50 nmol/L. Setting the specificity at a level of 95%, the cut-off for SalE_LN would be 20.50 nmol/L.

After overnight 1-mg dexamethasone suppression, the optimal cut-off for SalF_Dex was 0.85 nmol/L (Table 3). Nonetheless, these values should be interpreted with caution, since many SalF_Dex values in both the eucortisolism (87.2%) and hypercortisolism (19.0%) groups were below the detection limit of the LC-MS/MS assay and were consequently presumed to be equivalent to the lowest detection limit of 0.5 nmol/L.

After 1-mg overnight dexamethasone suppression, the optimal cut-off for SalE_Dex was 7.45 nmol/L. Setting the sensitivity at a level of 95%, the cut-off for SalE_Dex would be 3.25 nmol/L (Table 3).

The correlation between 0900h serum cortisol and salivary cortisol was 0.81 (P<0.01); and that between 0900h serum cortisol and salivary cortisone was 0.88 (P<0.01). The correlation between SerF_Dex and SalF_Dex was 0.90 (P<0.01); and that between SerF_Dex and SalE_Dex was 0.94 (P<0.01) [Fig 2].

All investigators who study Cushing’s syndrome are confronted with the conundrum of accurately diagnosing or excluding the condition with no gold standard test.1 In our study, in addition to the histological criterion, we considered it appropriate to include a set of biochemical criteria in which subjects with two positive results among the three commonly used tests—namely the SerF_Dex, the UFC, and the SalF_LN—were considered to have hypercortisolism. This is in agreement with the Endocrine Society Clinical Practice Guideline17 that recommends performing one or two other tests if one of these is abnormal; and if results from two different tests are concordant, to proceed with investigations to establish the cause of Cushing’s syndrome. One well-recognised contentious point in the interpretation of the SerF_Dex is the optimal cut-off: <140 nmol/L is a widely cited normal response, but can lead to false-negative results in up to 15% of subjects with Cushing’s syndrome.18 19 The more stringent cut-off of <50 nmol/L sacrifices specificity for sensitivity.20 21 In this study, we adopted a double cut-off as proposed by the European Society of Endocrinology Clinical Practice Guideline in collaboration with the European Network for the Study of Adrenal Tumors15; the rationale being that a more sensitive cut-off should be employed in those with a higher pretest probability of Cushing’s syndrome, such as the presence of an adrenal adenoma on imaging studies.22 A more specific cut-off can be employed in general to avoid overdiagnosis.

The loss of circadian rhythm with absence of a late-night cortisol nadir is a well-established feature of Cushing’s syndrome. Midnight serum cortisol is, however, difficult to obtain. When SalF_LN was shown to correlate well with serum cortisol levels, with sensitivity of 92%-100% and specificity of 93%-100% for the diagnosis of Cushing’s syndrome,17 it rapidly became one of the most popular tests in investigating endogenous hypercortisolism. In view of the theoretical advantages of salivary cortisone, we also attempted to explore the performance characteristics of SalE_LN. Our data showed that it had a good sensitivity of 94.7% and a specificity of 87.2% at the cut-off of 13.50 nmol/L, as measured by LC-MS/MS.

We could not compare the utility of SalF_LN with cortisone in this study, since SalF_LN was one of the criteria applied to define Cushing’s syndrome. Simultaneous measurement of salivary cortisol and salivary cortisone can nonetheless alert clinicians to certain caveats encountered when measuring salivary cortisol alone. When the salivary cortisol-to-cortisone ratio is exceptionally high, direct contamination of the oral sample by topical or oral hydrocortisone must be excluded. Ingestion of glycyrrhetinic acid (eg in licorice, carbenoxolone), which competitively inhibits 11β-HSD2, or rare cases of genetic 11β-HSD2 defect, can also lead to the same anomaly.

A number of other investigators have explored the utility of SalF_Dex in the diagnosis of Cushing’s syndrome. Apart from its convenience, salivary values are not affected by conditions that affect corticosteroid-binding globulin (CBG) or albumin levels, such as acute and chronic illness, pregnancy or oestrogen treatment, or genetic variants of CBG. A sensitivity varying between 97% and 100% and a specificity between 77% and 100% have been variously reported, with cut-off level varying between 1.7 nmol/L and 2 nmol/L.6 7 8 9 In the current study, we found that the sensitivity of SalF_Dex was only 76.2% at the optimal cut-off of 0.85 nmol/L. Although this discrepancy with other studies might be due to a number of factors, such as the means of defining normal ranges and the criteria for diagnosing Cushing’s syndrome, one important factor that is evident from our data is the method used for assaying salivary cortisol: some used electrochemiluminescence assay,9 others used radioimmunoassay6 7 8; but we measured SalF_Dex with LC-MS/MS.12 Unlike immunoassays, LC-MS/MS measurement of analytes is more specific, with less cross-reactivity among different cortisol precursors and metabolites.23 The concentration of SalF_Dex was very low: 19% of our patients with hypercortisolism and 87% of those with eucortisolism had SalF_Dex below the detection limit of 0.5 nmol/L, leading to uncertainty in establishing the cut-off, since all those with results of <0.5 nmol/L could only be considered to have salivary cortisol level equal to 0.5 nmol/L in the analysis. Immunoassays, by measuring other cortisol precursors or metabolites in varying degrees in addition, could have bypassed this problem. Other studies have also reported that SalF_LN has poorer diagnostic performance characteristics if measured by LC-MS/MS, in comparison with the less-specific immunoassays such as chemiluminescent assays or radioimmunoassays.24 25

Nevertheless, LC-MS/MS is analytically more superior and is expected to become the steroid assay of choice in the future.26 Values generated by studies using LC-MS/MS have greater inter-centre and long-term generalisability in view of the lack of assay-specific steroid cross-reactivity. Adoption of cut-offs generated by studies in which salivary cortisol was assayed using antibody-based methods into clinical practice is known to be problematic.27 Individual centres are often advised to generate their own references and cut-offs although this is often not feasible. In a meta-analysis on the use of SalF_LN for investigation of Cushing’s syndrome,25 the recommended cut-offs varied widely, from 3.59 to 15.17 nmol/L. To overcome the problem of lower performance characteristics due to low levels of salivary cortisol, instead of going for immunoassays, a better solution may be to measure salivary cortisone that is present in a much higher concentration than salivary cortisol. At a serum cortisol below 74 nmol/L, Debono et al28 showed that salivary cortisol could become undetectable by LC-MS/MS, while salivary cortisone was always detected. Similarly, our data showed that after dexamethasone suppression, when the salivary cortisol became too low to be measured with LC-MS/MS in many subjects, salivary cortisone could still be measured in all but one subject in the eucortisolism group.

Between salivary cortisol and salivary cortisone, this study showed that salivary cortisone would be the preferred test because it is present at a higher concentration in the saliva; and at comparable specificity levels, SalE_Dex appears to have better accuracy (as reflected by the higher AUC of the ROC curves), sensitivity, and negative LR than SalF_Dex.

Apart from the optimal cut-off, clinically it is often useful to have two cut-offs, one for ruling in a diagnosis (high specificity) and another one for excluding a diagnosis (high sensitivity), depending on clinicians’ preference. If we arbitrarily define an acceptable and useful cut-off as having a 95% level of either sensitivity or specificity, the two useful cut-offs for SalE_LN as derived from our study were 13.50 and 20.50 nmol/L, respectively; those for SalE_Dex were 3.25 nmol/L and 7.45 nmol/L, respectively.

Traditionally, in the algorithm for the workup for Cushing’s syndrome, late-night levels (serum or salivary cortisol) have been used for screening (excluding Cushing’s syndrome), whereas the post-dexamethasone level (serum cortisol) has been used for diagnosis (ruling in Cushing’s syndrome). When used as such, the cut-off of 13.50 nmol/L can be used for SalE_LN; whereas 7.45 nmol/L can be used for SalE_Dex. For the time being, SalE_Dex data can supplement serum cortisol measurement as a confirmatory test when concordant, or alert the clinician to the potential pitfalls with serum cortisol (eg variations in CBG levels) when discordant. With more experience, SalE_Dex may even ultimately replace the need to measure serum cortisol.

The strength of this study lies in the rigour with which a pre-specified protocol was adhered to. A high success rate of sample collection was achieved, with little missing data. Insufficient salivary volume collected in the Salivette tubes was the most common reason for unsuccessful salivary collection, because LC-MS/MS requires a larger saliva volume (100-250 µL) than immunoassay (40-50 µL).29 Two thirds of our study subjects were referred either because of an adrenal incidentaloma or common clinical conditions such as diabetes mellitus and hypertension, and had no clinical features of Cushing’s syndrome. This population was quite representative of cases referred to an endocrine centre for workup of Cushing’s syndrome.

A notable limitation of this study is the small number of subjects with hypercortisolism. The cut-off for the SerF_Dex was adopted from the literature rather than from data derived from our own healthy volunteers. The Endocrine Society Clinical Practice Guideline17 recommended two separate measurements of SalF_LN or UFC. Only one sample for each was collected in our study. Although we might have consequently missed some cases of episodic hypercortisolism, we assumed that if less than two out of the relatively sensitive tests were positive at the time of sampling, the subjects would likely be in a phase of normal cortisol secretion even if they had episodic Cushing’s syndrome.

Our study showed that salivary cortisone can become the analyte of choice for investigating Cushing’s syndrome in the era of LC-MS/MS. Our data suggest using 13.50 nmol/L for SalE_LN, and either 7.45 nmol/L (more specific) or 3.25 nmol/L (more sensitive) for SalE_Dex, as cut-offs.

All authors have disclosed no conflicts of interest.

1. Stewart PM. Is subclinical Cushing’s syndrome an entity or a statistical fallout from diagnostic testing? Consensus surrounding the diagnosis is required before optimal treatment can be defined. J Clin Endocrinol Metab 2010;95:2618-20. Crossref

2. Ferraù F, Korbonits M. Metabolic comorbidities in Cushing’s syndrome. Eur J Endocrinol 2015;173:M133-57. Crossref

3. Viardot A, Huber P, Puder JJ, Zulewski H, Keller U, Müller B. Reproducibility of nighttime salivary cortisol and its use in the diagnosis of hypercortisolism compared with urinary free cortisol and overnight dexamethasone suppression test. J Clin Endocrinol Metab 2005;90:5730-6. Crossref

4. Graham UM, Hunter SJ, McDonnell M, Mullan KR, Atkinson AB. A comparison of the use of urinary cortisol to creatinine ratios and nocturnal salivary cortisol in the evaluation of cyclicity in patients with Cushing’s syndrome. J Clin Endocrinol Metab 2013;98:E72-6. Crossref

5. Vining RF, McGinley RA, Maksvytis JJ, Ho KY. Salivary cortisol: a better measure of adrenal cortical function than serum cortisol. Ann Clin Biochem 1983;20(Pt 6):329-35. Crossref

6. Barrou Z, Guiban D, Maroufi A, et al. Overnight dexamethasone suppression test: comparison of plasma and salivary cortisol measurement for the screening of Cushing’s syndrome. Eur J Endocrinol 1996;134:93-6. Crossref

7. Cardoso EM, Arregger AL, Tumilasci OR, Contreras LN. Diagnostic value of salivary cortisol in Cushing’s syndrome (CS). Clin Endocrinol (Oxf) 2009;70:516-21.

8. Castro M, Elias PC, Quidute AR, Halah FP, Moreira AC. Out-patient screening for Cushing’s syndrome: the sensitivity of the combination of circadian rhythm and overnight dexamethasone suppression salivary cortisol tests. J Clin Endocrinol Metab 1999;84:878-82. Crossref

9. Deutschbein T, Broecker-Preuss M, Flitsch J, et al. Salivary cortisol as a diagnostic tool for Cushing’s syndrome and adrenal insufficiency: improved screening by an automatic immunoassay. Eur J Endocrinol 2012;166:613-8. Crossref

10. Perogamvros I, Keevil BG, Ray DW, Trainer PJ. Salivary cortisone is a potential biomarker for serum free cortisol. J Clin Endocrinol Metab 2010;95:4951-8. Crossref

11. Wood P. Salivary steroid assays—research or routine? Ann Clin Biochem 2009;46(Pt 3):183-96. Crossref

12. Antonelli G, Ceccato F, Artusi C, Marinova M, Plebani M. Salivary cortisol and cortisone by LC-MS/MS: validation, reference intervals and diagnostic accuracy in Cushing’s syndrome. Clin Chim Acta 2015;451(Pt B):247-51.

13. Smith RE, Maguire JA, Stein-Oakley AN, et al. Localization of 11 beta-hydroxysteroid dehydrogenase type II in human epithelial tissues. J Clin Endocrinol Metab 1996;81:3244-8. Crossref

14. Perogamvros I, Owen LJ, Newell-Price J, Ray DW, Trainer PJ, Keevil BG. Simultaneous measurement of cortisol and cortisone in human saliva using liquid chromatography-tandem mass spectrometry: application in basal and stimulated conditions. J Chromatogr B Analyt Technol Biomed Life Sci 2009;877:3771-5. Crossref

15. Fassnacht M, Arlt W, Bancos I, et al. Management of adrenal incidentalomas: European Society of Endocrinology Clinical Practice Guideline in collaboration with the European Network for the Study of Adrenal Tumors. Eur J Endocrinol 2016;175:G1-G34. Crossref

16. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143:29-36. Crossref

17. Nieman LK, Biller BM, Findling JW, et al. The diagnosis of Cushing’s syndrome: an Endocrine Society Clinical Practice Guideline. J Clin Endocrinol Metab 2008;93:1526-40. Crossref

18. Findling JW, Raff H, Aron DC. The low-dose dexamethasone suppression test: a reevaluation in patients with Cushing’s syndrome. J Clin Endocrinol Metab 2004;89:1222-6. Crossref

19. Görges R, Knappe G, Gerl H, Ventz M, Stahl F. Diagnosis of Cushing’s syndrome: re-evaluation of midnight plasma cortisol vs urinary free cortisol and low-dose dexamethasone suppression test in a large patient group. J Endocrinol Invest 1999;22:241-9. Crossref

20. Ceccato F, Barbot M, Zilio M, et al. Screening tests for Cushing’s syndrome: urinary free cortisol role measured by LC-MS/MS. J Clin Endocrinol Metab 2015;100:3856-61. Crossref

21. Nieman LK. Cushing’s syndrome: update on signs, symptoms and biochemical screening. Eur J Endocrinol 2015;173:M33-8. Crossref

22. Lopez D, Luque-Fernandez MA, Steele A, Adler GK, Turchin A, Vaidya A. “Nonfunctional” adrenal tumors and the risk for incident diabetes and cardiovascular outcomes: a cohort study. Ann Intern Med 2016;165:533-42. Crossref

23. Raff H, Auchus RJ, Findling JW, Nieman LK. Urine free cortisol in the diagnosis of Cushing’s syndrome: is it worth doing and, if so, how? J Clin Endocrinol Metab 2015;100:395-7. Crossref

24. Erickson D, Singh RJ, Sathananthan A, Vella A, Bryant SC. Late-night salivary cortisol for diagnosis of Cushing’s syndrome by liquid chromatography/tandem mass spectrometry assay. Clin Endocrinol (Oxf) 2012;76:467-72. Crossref

25. Carroll T, Raff H, Findling JW. Late-night salivary cortisol for the diagnosis of Cushing syndrome: a meta-analysis. Endocr Pract 2009;15:335-42. Crossref

26. Handelsman DJ, Wartofsky L. Requirement for mass spectrometry sex steroid assays in the Journal of Clinical Endocrinology and Metabolism. J Clin Endocrinol Metab 2013;98:3971-3. Crossref

27. Beko G, Varga I, Glaz E, et al. Cutoff values of midnight salivary cortisol for the diagnosis of overt hypercortisolism are highly influenced by methods. Clin Chim Acta 2010;411:364-7. Crossref

28. Debono M, Harrison RF, Whitaker MJ, et al. Salivary cortisone reflects cortisol exposure under physiological conditions and after hydrocortisone. J Clin Endocrinol Metab 2016;101:1469-77. Crossref

29. Raff H. Cushing’s syndrome: diagnosis and surveillance using salivary cortisol. Pituitary 2012;15:64-70. Crossref

The human gonads, both the ovaries and testes, are sensitive organs susceptible to injury by disease, medications, and chemotherapy and radiation for the treatment of cancers and other medical conditions including autoimmune diseases such as systemic lupus erythematosus and haematological diseases.1 2 3 Individuals who survive may later consider starting a family, yet by this time they often face problems of gonadal injury and ageing. If their fertility can be preserved before such treatment is performed, especially at a young age, individuals will be able to retain or regain their fertility after completion of treatment.

Current advances in reproductive technology have enabled fertility to be retained by preservation of gonadal function such that gametes as well as hormones continue to be produced despite damage inflicted by gonadotoxic treatment. Fertility preservation methods include fertility-sparing surgery, radiation shielding, and gonadotropin-releasing hormone agonists for gonadal suppression during chemotherapy. In addition, assisted reproductive technology—including intracytoplasmic sperm insemination; and oocyte, embryo and ovarian tissue cryopreservation—have expanded fertility preservation options that can now be applied to a broader spectrum of patients including those who are pre-pubertal, and those with insufficient time prior to initiation of gonadotoxic treatment.4 5 6

Although any adverse effects of treatment on fertility should have been discussed by clinicians before treatment, up to half of the patients are not referred to fertility specialists for fertility preservation.7 To the best of our knowledge, there is no local literature on the awareness of, attitude towards, and knowledge about fertility preservation among clinicians in Hong Kong. We therefore conducted a questionnaire survey to address this issue.

This was a cross-sectional survey to evaluate the awareness of, attitude towards, and knowledge about fertility preservation among local clinicians in Hong Kong. The study was conducted between June 2016 and December 2016. Ethical approval for the study was obtained from the institutional Survey and Behavioural Research Ethics Committee.

Eligible subjects were identified from the Specialist Register of Medical Council of Hong Kong who were clinicians worked in public hospitals and specialised in the field of Clinical Oncology, Haematology, Obstetrics and Gynaecology, Paediatrics, and Surgery. Potential candidates were selected by convenience sampling from each specialty from various hospitals and their work addresses identified via the electronic staff directory or organisational chart provided by the Hospital Authority intranet. The study questionnaire was mailed to them internally.

The self-administered questionnaire included a brief explanation of the survey. If the subject agreed to participate, they were asked to complete the questionnaire and return it in the stamped addressed envelope. The questionnaires were completed and returned anonymously.

The questionnaire comprised 29 items in two parts. The first part included questions about the baseline demographics and specialty of the participants. Their views on the demand for the fertility preservation service, factors they considered when making a decision about fertility preservation, and the difficulties encountered in discussing fertility issues with their patients were examined. Practical questions about the potential costs and the need for a dedicated fertility preservation clinic were also addressed.

The SPSS (Windows version 20.0; IBM Corp, Armonk [NY], US) was used for data entry and analysis. Demographic data were summarised by means, medians, and percentages. The Chi squared test (χ² test) was used for categorical data such as comparing the awareness of fertility preservation among different specialties, cancer type, and demographic background. Student’s t test (t test) was used for continuous variables of age and years of practice. Results with a P value of <0.05 were considered statistically significant.

Of the 467 questionnaires sent to a convenient sample of clinicians, 10 were returned unopened because of an outdated work address. A total of 167 questionnaires of the remaining 457 questionnaires were returned, giving an overall response rate of 36.5%. The response rates for specific specialties were: 55.3% (68/123) for obstetricians and gynaecologists, 37.5% (48/128) for surgeons (general/breast/urology), 18.5% (22/119) for paediatricians, and 16.5% (16/97) for haematologists or clinical oncologists. Table 1 summarises the baseline demographics of the respondents. Some of the respondents did not answer all questions, hence the denominators of each response are stated.

Up to 85.0% (142/167) of respondents cared for cancer patients in their daily practice and 76.0% (127/167) dealt with treatments that may threaten fertility. The most commonly encountered cancers were gynaecological cancer (50.0%, 71/142), followed by urological cancer (25.4%, 36/142), haematological cancer (20.4%, 29/142), neurological cancer (19.7%, 28/142), musculoskeletal cancer (18.3%, 26/142), gastrointestinal cancer (16.2%, 23/142), and others (6.3%, 9/142).

Only 45.6% (73/160) of the respondents were familiar with fertility preservation. The three most familiar means were sperm freezing (66.3%, 108/163), followed by oocyte freezing (65.0%, 106/163) and embryo freezing (50.3%, 82/163). Table 2 shows the awareness of various fertility preservation strategies among clinicians from different specialties.

Nevertheless, 68.3% (112/164) of respondents had never referred a patient for fertility preservation. Among the 52 respondents who had, 88.5% (46/52) had referred fewer than five patients and 11.5% had referred more than five patients over the past 12 months. Sperm cryopreservation was the most commonly referred fertility preservation method. There was no significant association of the demographic background of respondents in terms of age (P=0.334), gender (P=0.325), marital status (P=0.060), presence of any children (P=0.574), or practice setting (P=0.749) with awareness or frequency of referral for fertility preservation. Up to 90.7% (146/161) would consider referral of a patient to a fertility specialist for fertility preservation if it delayed treatment by 1 week, 83.2% (134/161) if the delay was <2 weeks, 41.6% (67/161) for <4 weeks, and 6.2% (10/161) for <8 weeks.

Table 3 shows the responses to questions about fertility preservation. Up to 76.5% (117/153) of the respondents agree that fertility preservation should be available as a public service. The top five difficulties encountered by clinicians in discussing fertility preservation were: no time before commencement of gonadotoxic treatment (60.6%, 97/160), high risk of cancer recurrence (53.8%, 86/160) or poor prognosis, financial constraints (46.9%, 75/160), treating the cancer as top priority (38.8%, 62/160), and not being aware of any place or person to whom their patients could be referred to (35.0%, 56/160).

Gonadotoxic treatments for cancer, especially those requiring chemotherapy with alkylating agents and total body irradiation or pelvic/whole-body radiation, have a significant negative impact on ovarian and testicular function.1 These impacts may be irreversible depending on the patient’s age, total dose administered, and gonadal reserve at the time of treatment.

Fertility preservation has gained increasing attention worldwide over the past decade as treatment advances result in more and more survivors of childhood cancers and adult malignancies who are expected to lead a normal life and to start a family of their own.1

Our study revealed several important findings. First, it showed a rather low awareness of fertility preservation among our respondents. Most agreed that their patients should be referred for fertility preservation even if it meant a delay in their treatment. Although up to three quarters of respondents dealt with treatment that might impair fertility, less than half were familiar with fertility preservation. Our previous study showed significant underutilisation of a sperm cryopreservation service over the past two decades.8 There is an imperative need to provide better education and campaigns to raise awareness about various options for fertility preservation available in Hong Kong.

Second, our study evaluated the difficulties or barriers encountered by clinicians in referring patients for fertility preservation. Similar to previous studies, a high risk of disease recurrence and poor prognosis discouraged discussion about future fertility.9 10 11 More than half of the respondents also expressed insufficient time for fertility preservation procedures before initiation of gonadotoxic treatment. Nonetheless sperm cryopreservation is a simple and effective method of preserving fertility for male patients who need to produce only a semen sample by masturbation for cryopreservation at any time before initiation of gonadotoxic treatment.8 In female patients, fertility preservation is slightly more complicated and time-consuming. Ovarian stimulation for oocyte or embryo cryopreservation takes at least 8 to 12 days although the introduction of random-start protocols for ovarian stimulation and ovarian tissue cryopreservation now provide a new option for those with insufficient time and for pre-pubertal adolescents.5 6 12 13 Early referral to a fertility specialist at the time of diagnosis of disease and prior to treatment commencement is the key to maximising the success of fertility preservation and allows a greater window of opportunity for preserving fertility.14 Again, this highlighted the need for training and education of clinical practitioners in the most updated advances in assisted reproductive technology, especially in specialties other than obstetrics and gynaecology.

Third, almost all respondents agreed there was a need for a dedicated clinic or referral centre. Most suggested two centres, catering to both private and public patients. No such referral centre is currently available in Hong Kong. An important prerequisite is a quick and efficient system whereby patients can be referred for fertility preservation counselling by a fertility specialist as soon as their diagnosis of cancer is made.15 Moreover, proper regulations and guidelines about fertility preservation should be established and communicated to the public and clinicians. Printed information about the effect of cancer treatment on fertility and the options for fertility preservation techniques, including both established and experimental, should be available for all clinicians to hand out to their patients. A 24-hour hotline should be set up and contact addresses disseminated widely on websites or to clinicians who care for patients with cancer.

Fourth, financial constraints should be addressed. Cryopreservation of gametes and embryos is expensive and is currently only available in Hong Kong as a private service. Government and non-governmental organisations should consider funding this in selected patients. Up to 76.5% of our respondents agreed that fertility preservation should be provided as a public service.

In addition, there appeared to be varying levels of awareness among clinicians from different specialties about fertility preservation techniques. Different specialists may be more or less exposed to the most up-to-date trends in the field of assisted reproductive technology. Our data were not sufficiently representative to explore this issue. Further studies are required to evaluate this.

Our study is limited by its small sample size and low response rate. Ideally, all clinicians from both public and private sectors of all specialties should be included but this would be costly and impractical. Our study included a higher proportion of clinicians from university-affiliated hospitals and this might have added additional self-selection bias to the study as they were more willing to participate in research. In addition, clinicians with an interest in this area may have been more likely to respond to this study. Potential candidates were sampled by convenience from each specialty from various hospitals and might not have represented the views of all clinicians. Caution should be exercised when making generalisations about these data from a sample group that was self-selected. Nonetheless this is the first study to evaluate the awareness of, attitude towards, and knowledge about fertility preservation among clinicians in Hong Kong. It provides important information that can be applied in setting up a fertility preservation centre and in the design of training modules and educational materials for clinical practitioners.

Reassuringly, our studies show an overall encouraging positive attitude among local clinicians towards fertility preservation, with the majority wanting to know more. Knowledge about fertility preservation techniques is insufficient. There is a need to improve awareness of and referral for this service. As the field of fertility preservation continues to grow, it is important to include the topic of fertility preservation in the curriculum of our medical schools to increase the knowledge and awareness of our future clinicians. Seminars, workshops, and conferences for those interested in this field should be regularly arranged. Fundraising campaigns and grants for research in this field should be encouraged. A multidisciplinary team and dedicated centre with an efficient referral system should be set up as soon as possible to provide fertility risk assessment and counselling for patients. Further studies are required to explore how fertility concerns are being addressed during the management of serious medical conditions, especially cancer care, and how clinicians can communicate with cancer patients about the options for fertility preservation.

We would like to thank Miss Elaine Yee-lee Ng, Miss Sze-yan Lo, and Mr Ka-chun Keung for data collection and entry. We would also like to thank all the clinicians who participated in this study.

All authors have disclosed no conflicts of interest.

1. Rodriguez-Wallberg KA, Oktay K. Fertility preservation during cancer treatment: clinical guidelines. Cancer Manag Res 2014;6:105-17.

2. Falcone T, Bedaiwy MA. Fertility preservation and pregnancy outcome after malignancy. Curr Opin Obstet Gynecol 2005;17:21-6. Crossref

3. Loren AW, Mangu PB, Beck LN, et al. Fertility preservation for patients with cancer: American Society of Clinical Oncology clinical practice guideline update. J Clin Oncol 2013;31:2500-10. Crossref

4. Ethics Committee of American Society for Reproductive Medicine. Fertility preservation and reproduction in patients facing gonadotoxic therapies: a committee opinion. Fertil Steril 2013;100:1224-31. Crossref

5. Dolmans MM, Jadoul P, Gilliaux S, et al. A review of 15 years of ovarian tissue bank activities. J Assist Reprod Genet 2013;30:305-14. Crossref

6. Donnez J, Martinez-Madrid B, Jadoul P, Van Langendonckt A, Demylle D, Dolmans MM. Ovarian tissue cryopreservation and transplantation: a review. Hum Reprod Update 2006;12:519-35. Crossref

7. Forman EJ, Anders CK, Behera MA. A nationwide survey of oncologists regarding treatment-related infertility and fertility preservation in female cancer patients. Fertil Steril 2010;94:1652-6. Crossref

8. Chung JP, Haines CJ, Kong GW. Sperm cryopreservation for Chinese male cancer patients: a 17-year retrospective analysis in an assisted reproductive unit in Hong Kong. Hong Kong Med J 2013;19:525-30. 

9. Shimizu C, Bando H, Kato T, Mizota Y, Yamamoto S, Fujiwara Y. Physicians’ knowledge, attitude, and behavior regarding fertility issues for young breast cancer patients: a national survey for breast care specialists. Breast Cancer 2013;20:230-40. Crossref

10. Arafa MA, Rabah DM. Attitudes and practices of oncologists toward fertility preservation. J Pediatr Hematol Oncol 2011;33:203-7. Crossref

11. Collins IM, Fay L, Kennedy MJ. Strategies for fertility preservation after chemotherapy: awareness among Irish cancer specialists. Ir Med J 2011;104:6-9. 

12. Cakmak H, Rosen MP. Random-start ovarian stimulation in patients with cancer. Curr Opin Obstet Gynecol 2015;27:215-21. Crossref

13. Rashidi BH, Tehrani ES, Ghaffari F. Ovarian stimulation for emergency fertility preservation in cancer patients: a case series study. Gynecol Oncol Rep 2014;10:19-21. Crossref

14. Yee S, Fuller-Thomson E, Lau A, Greenblatt EM. Fertility preservation practices among Ontario oncologists. J Cancer Educ 2012;27:362-8. Crossref

15. Ghazeeri G, Zebian D, Nassar AH, et al. Knowledge, attitudes and awareness regarding fertility preservation among oncologists and clinical practitioners in Lebanon. Hum Fertil (Camb) 2016;19:127-33. Crossref

Professional burnout can be one of the possible consequences of chronic occupational stress.1 Burnout has been described as a gradual erosion of the person and comprises three characteristics.2 First, the individuals are exhausted, either mentally or emotionally.2 They feel drained, tired with insufficient energy, and are unable to cope.3 Second, they may have a negative, indifferent, or cynical attitude towards patients, clients, or colleagues.2 They feel numb about work and distance themselves emotionally.3 Finally, they also tend to feel dissatisfied with their own performance and evaluate themselves negatively.2 They find difficulty in concentrating on their work and caring about their families.3 The consequences of burnout can be very serious. Maslach and Jackson4 found that the quality of care and service provided by individuals who have burnout syndrome may be substandard. Burnout has also been found to be associated with job turnover, absenteeism, low morale, and personal dysfunction.3 Maslach and her colleague5 6 have studied the burnout syndrome for many years and devised a measurement tool—the Maslach Burnout Inventory—Human Services Survey (MBI-HSS).4 7 Although some studies of burnout have been conducted among dentists using the MBI-HSS,1 8 9 10 most studies on stress and burnout in health professions have focused on physicians11 12 13and nurses.14 15

In 2001, a survey was conducted to investigate the sources of occupation stress among dentists in Hong Kong.16 There has been no published study on burnout among dentists in Hong Kong since then. We conducted a cross-sectional study to investigate the burnout among dentists in Hong Kong and the association between occupational stress and burnout. The objectives of our study were to identify the occupational stressors and the prevalence of burnout among dentists in Hong Kong and their association.

In October 2015, there were 2173 locally registered dentists working in Hong Kong according to the Dental Council of Hong Kong.17 In order to achieve the width of the 95% confidence interval (CI) calculated from the sample proportion estimates of burnout outcomes no wider than ± 5%, a sample of 384 dentists was needed (using a conservative assumption of sample proportions to be 50%). Taking into account the low response rate (<40%) achieved by previous surveys in Hong Kong, 50% of the dentists were systematically randomly sampled (with a random start of ‘2’ generated from Excel) from the published list of registered dentists according to their family names and a sample size of 1086 was chosen. The questionnaires were mailed in sealed envelopes in early November 2015 along with a stamped, self-addressed envelope. The participants were assured of confidentiality. No names and addresses were asked for and no codes were marked on the questionnaires or return envelopes. Questionnaires for selected government dentists were sent collectively to the Department of Health and delivered internally. Reminders were sent to all selected private dentists in late November or early December 2015. Participants did not receive any specific incentive to complete this survey.

The questionnaire assessed the sources of occupational stress, stress-coping strategies, effects of stress on work, level of burnout, and socio-demographic characteristics of the study subjects. There were 33 occupational stressors for identifying the sources of stress; 26 of which were proposed by Cooper et al.18 An additional seven stressors were proposed by Waddington.19 Participants were asked to rate the level of stress they experienced on a 5-point scale ranging from ‘1 = No stress’ to ‘5 = A great deal of stress’. The 33 stressors were grouped into five domains: patient-related, time-related, income-related, job-related, and staff-related or technically related (Box). Based on the work by Cooper et al,18 10 stress-coping strategies of dentists were included. Participants were asked how often they would use these stress-coping strategies with responses categorised on a 5-point scale ranging from ‘1 = Never’ to ‘5 = Always’ (Box). Based on the work of Kopec and Esdaile,20 five effects of occupational stress on work were included with responses rated on a 4-point scale ranging from ‘1 = Not at all’ to ‘4 = A lot’ (Box). Of note, MBI-HSS,7 which has been recognised as the leading measure of burnout, was also included and consisted of 22 items with responses rated on a 7-point scale ranging from ‘0 = Never’ to ‘6 = Every day’. The MBI-HSS addressed three general scales: emotional exhaustion (EE; 9 items with score range of 0-54) measuring the feelings of being emotionally overextended and exhausted by one’s work; depersonalisation (DP; 5 items with score range of 0-30) measuring the unfeeling and impersonal response towards recipients of one’s service, care treatment, or instruction; and personal accomplishment (PA; 8 items with score range of 0-48) measuring the feelings of competence and successful achievement in one’s work. Personal information about the participants was also collected and included their age, gender, years of practice, type of practice, location, hours of work per week, marital status, working status of spouse, number of children, religious beliefs, whether the participant had postgraduate qualifications, and whether the participant was a specialist.

The Institutional Review Board of The University of Hong Kong/Hospital Authority Hong Kong West Cluster (HKU/HA HKW IRB, reference number: UW 15-524) approved the study prior to distribution of the questionnaires.

The mean score for each stressor group was calculated and ranked. Three subscale scores of MBI-HSS (EE, DP, and PA) were calculated and subjects categorised into one of the three groups—high, moderate, or low level of burnout. Those with EE score of ≥27 were considered to have high EE level, 17-26 moderate and 0-16 low level. Those with DP score of ≥13 were considered to have high DP level, 7-12 moderate and 0-6 low level. Those with PA score of ≤31 were considered to have low PA level, 32-38 moderate and ≥39 high level. Subjects with high EE, high DP, and low PA levels simultaneously were considered to have a high level of overall burnout. Others were considered to have low overall burnout. Of the results, 95% CIs were computed as well.

The relationships between EE, DP and PA scores, and the set of 17 independent variables (12 demographic variables and the five stressor group scores) were analysed by analysis of covariance (ANCOVA). A forward selection method was adopted. In this approach, independent variables were added into the model one at a time. In each step, each variable that was not already in the model was tested for inclusion in the model. The most significant of these variables was added to the model. Finally, only significant variables were selected in the final model.

The relationship between levels of overall burnout (high and low) and the set of 17 independent variables was analysed by multiple logistic regression. A forward selection method was adopted as well. All analyses were performed using the SPSS (Windows version 22.0; IBM Corp, Armonk [NY], US) and the level of significance was set at 0.05.

A total of 1086 questionnaires were sent to the randomly selected dentists. Reminders were sent to the selected private dentists 2 to 4 weeks later. We received 301 completed questionnaires and 22 questionnaires returned undelivered by the post office, resulting in a response rate of 28.3% (301/[1086–22] x 100%).

The demographics of the respondents are shown in Table 1. Approximately 65% of the respondents were male and 79% worked in the private sector that also included non-governmental organisations. Approximately 46% of dentists had acquired one or more postgraduate qualifications. More than 86% of the respondents were general dental practitioners. Over a quarter (29%) had never married and more than half (54%) claimed to have no religious beliefs.

The top 10 stressors according to the percentage distribution of the respondents’ ratings ‘4’ or ‘5’ (considered to be high level) are shown in Figure 1. Six of the top 10 highest ranked stressors (and actually the top five highest ranked) were patient-related stressors and three of the 10 highest ranked stressors were time-related. The patient-related stressors group had the highest mean (± standard deviation [SD]) score (3.4 ± 0.8), followed by the time-related stressors group (3.1 ± 0.8), staff-related or technically related stressors group (2.6 ± 0.6), income-related stressors group (2.5 ± 0.7), and job-related stressors group (2.5 ± 0.7). Female dentists had a higher mean score than male dentists for patient-related, job-related, and staff-/technically related stressors (3.5 vs 3.3, 2.6 vs 2.4, and 2.7 vs 2.5, respectively; all P<0.05). Dentists with more than 20 years of practice (3.3), who held postgraduate qualifications (3.2), or who had completed specialist training (3.0) had a lower mean score for patient-related stressors than those with less than 20 years of practice (3.4, 3.5, and 3.4 respectively; all P<0.05).

The highest ranked strategy (with ratings ‘4’ or ‘5’) to cope with occupational stress was ‘try to control one’s own working situation / condition’ (59.0%), followed by ‘pursue outside interests’ (58.6%), ‘avoid the stressful situation’ (51.7%), and ‘take exercise’ (51.0%). Other less-commonly used strategies (<50%) were ‘re-interpret problem positively’, ‘seek support and advice’, ‘social gathering’, ‘devote yourself more to your work’, ‘take medication’, and ‘others’.

The highest ranked effect (with ratings ‘3’ or ‘4’) of occupational stress on work was to ‘cut down on the amount of extra work or overtime’ (41.7%) and ‘take more frequent / longer breaks’ (30.7%). Fewer than 30% of respondents reported ‘less satisfaction with your job’, ‘work more slowly’, and ‘less able to concentrate on your work’ as the effects of occupational stress on work.

The mean (± SD) score for EE, DP and PA and the level of burnout are shown in Figure 2. Approximately a quarter of the respondents had a high EE level (25.4%; 95% CI, 20.5%-30.3%) while 17.2% (95% CI, 12.9%-21.5%) and 39.0% (95% CI, 33.5%-44.5%) had high DP and low PA level, respectively. Nonetheless only 7.0% (95% CI, 4.1%-9.9%) of respondents had a high level of overall burnout (ie high EE, high DP, and low PA).

The relationship between EE, DP and PA scores, five stressor groups, and the 12 demographic variables was analysed by ANCOVA. The final models are shown in Table 2. For EE, among the 17 independent variables investigated, only three significant variables (P<0.05) were selected in the final model. The dentists whose spouses were not working had a higher mean EE score by 4.15 compared with those whose spouses were working (P=0.024). Those with higher scores in the time-related stressors group had higher EE scores. For every one unit increase in time-related stressors group score, the mean EE score increased by 4.51 (P<0.001). Finally, those with higher scores in the job-related stressors group also had higher EE scores; for every one unit increase in job-related stressors group score, the mean EE score increased by 4.54 (P<0.001). For DP, only two significant variables (P<0.05) were selected in the final model. Private dentists who worked in residential areas had higher mean DP scores by 2.51 compared with those private dentists who worked in commercial areas (P=0.005). Also, those with higher scores in the job-related stressors group also had higher DP scores; for every one unit increase in job-related stressors group score, the mean DP score increased by 2.35 (P<0.001). For PA, only one significant variable (P<0.05) was selected in the final model. Those with higher scores in income-related stressors group were found to have lower PA scores. For every one unit increase in job-related stressors group score, the mean PA score decreased by 3.40 (P<0.001).

The relationships between levels of overall burnout (high and low), five stressors groups, and the demographic variables were analysed by multiple logistic regression. The results are shown in Table 3; only two significant variables (P<0.05) were selected. The chance of having a high level of overall burnout among dentists with no postgraduate qualifications was 5.08 times higher (95% CI, 1.09-23.61) than those with postgraduate qualifications (P=0.038). Also, for every one unit increase in job-related stressors group score, the odds of having a high level of overall burnout was 3.74 times as likely (95% CI, 1.77-7.87; P=0.001).

This survey revealed that the highest ranked stressors were patient-related. Several stress-coping strategies and effects of occupational stress on work were identified. Only 7% of dentists had a high level of overall burnout. Dentists without postgraduate qualifications and higher job-related stressor scores were more likely to have a high level of overall burnout.

Comparison of these results with those of a previous study published in 200116 reveals that the situation in Hong Kong has changed little over 15 years. In both surveys, six of the top 10 stressors were patient-related, three were time-related, and one was income-related. There was, however, an increase in the percentage distributions of the stressors, indicating an increased stress level for dentists in Hong Kong in the past decade. This may be related to increasing demands and expectations of patients. Patients can now find much information on the internet and ask many questions based on the information they may find. This is a novel challenge for dentists. Moreover, the property rents in Hong Kong have increased much over the last 10 years with a consequent increase in the overhead costs of running a dental clinic. As a result, dentists need to work harder and see more patients to generate sufficient income to cover their costs.

Female dentists had higher mean scores than male dentists for patient-related, job-related, and staff-/technically related stressors. This may be because they are taking care of their own family as well as working as a dentist. Dentists with more experience, higher postgraduate qualifications, and specialist training had lower mean scores in the patient-related stressors group. This may imply that dentists with more competent skills and knowledge were less stressed.

In the current and 200116 studies, the most prevalent method of coping with stress was ‘try to control one’s own working situation / condition. Most dentists worked on their own and were self-reliant, even when they were facing stress.

Higher EE score was significantly associated with those dentists whose spouses were not working and those with higher mean scores in time-related and job-related stressors groups. The financial pressure on a dentist may be lessened if the spouse works and this may have contributed to the lower EE score. Higher DP score was significantly associated with working in a residential area and higher mean score in job-related stressors group. This may be due to differing patient profiles in residential areas. Those who seek dental care from dentists in residential areas may be less educated, and less willing or less able to pay for dental treatment. Thus, the dentists have to spend more time with each patient but accrue less income with a consequent higher DP score. Lower PA score was significantly associated with those dentists with lower mean score in income-related stressors group. One of the rewards for a dentist is to earn a living by providing a professional dental service. If a dentist has difficulty in earning, he may have a lower sense of PA.

A high level of overall burnout was significantly associated with a higher mean score in job-related stressors group and no postgraduate qualifications. Those with postgraduate qualifications may have better knowledge, technique, and communication skills to deal with patients. This may have contributed to lower overall burnout.

The mean scores of the three subscales (EE, DP, and PA) of MBI of studies in Korea,8 Northern Ireland,9 UK,10 Dutch,21 and Hong Kong doctors11 are shown in Figure 3 for comparison. Generally speaking, the mean scores of EE and DP of dentists in Hong Kong were lower than those in other countries while mean PA score was in between. It is possible that Hong Kong dentists can tolerate stress better. In Hong Kong, the competition to study dentistry is very strong. A secondary school student must perform very well in public examinations to be admitted to the only dental school in Hong Kong. Locally trained students are used to this kind of keen competition. In Hong Kong, lower mean EE and DP scores and higher mean PA scores among dentists suggest that they are not as stressful as medical doctors. This may be due to the different nature of their jobs. Most dentists in Hong Kong practise in the private sector and seldom need to be on-call. In comparison, a lot of Hong Kong medical doctors work in public hospitals and need to be on-call. Moreover, the public have high expectations of the public health care service and this may contribute to the higher level of burnout among Hong Kong doctors.

The level of burnout of the current study was compared with that reported in Korea8 and the UK.10 The percentage distribution of high EE (Hong Kong 25.4%, Korea 41.2%, and UK 42.2%) and DP (Hong Kong 17.2%, Korea 55.9%, and UK 19.5%) level of Hong Kong dentists was lower compared with that of Korean and UK dentists. Hong Kong dentists, however, felt a lower sense of PA as reflected by the higher percentage of low PA (Hong Kong 39.0%, Korea 31.5%, and UK 31.9%). This suggests that the dentists in Hong Kong had lower job satisfaction. Although the level of overall burnout of Hong Kong dentists was less than that in other countries, this issue should not be overlooked. Dentistry is quite a lonely profession that requires an individual to work alone most of the time. Peer support and sharing is important for the development of dentistry in Hong Kong.

The response rate of the current study was low (<30%) although it was similar to other studies of health care professionals in Hong Kong. The selected dentists might be very busy and might not have time to complete the questionnaires. Others might not have been interested. Those dentists with a high level of burnout might have felt the questions too sensitive and thus been unwilling to participate. For those who participated, as stress and burnout were self-reported, there might have been information bias because of social desirability and thus our results might underestimate the level of stress and burnout among dentists in Hong Kong. As in all cross-sectional surveys, a causal relationship could not be established. With the small sample size, there was a possible lack of statistical power for the multiple logistic regression as evidenced by the wide 95% CI of the odds ratio. In order to increase the response rate of future surveys, we may consider phoning each selected dentist to check if they have received the questionnaires and to invite them to participate. We may also consider using the internet for data collection.

Dentists with a higher stress level may consider working less to relieve pressure: approximately one quarter of dentists worked more than 50 hours a week, which contributed to the time-related stress. Dentists are advised to update their knowledge and skills to equip them for the increasing expectations and challenges from patients and society.

Patient-related stressors are the top occupational stressors experienced by dentists in Hong Kong. Nonetheless a small proportion of dentists have high overall burnout. There was a positive association between occupational stress and level of burnout.

We would like to thank the staff of the Faculty of Dentistry, The University of Hong Kong for their assistance in this study. We would also like to thank all the dentists who participated in the survey.

All authors have disclosed no conflicts of interest.

1. Gorter RC, Albrecht G, Hoogstraten J, Eijkman MA. Work place characteristics, work stress and burnout among Dutch dentists. Eur J Oral Sci 1998;106:999-1005. Crossref

2. Rada RE, Johnson-Leong C. Stress, burnout, anxiety and depression among dentists. J Am Dent Assoc 2004;135:788-94. Crossref

3. US National Library of Medicine. Depression: what is burnout? Available from: https://www.ncbi.nlm.nih.gov/pubmedhealth/PMH0072470/. Accessed 15 Mar 2017.

4. Maslach C, Jackson SE. Maslach burnout inventory manual. Mountain View, CA: Consulting Psychologists Press; 1981.

5. Maslach C. Burned-out. Can J Psychiatr Nurs 1979;20:5-9.

6. Maslach C, Jackson SE. The measurement of experienced burnout. J Organ Behav 1981;2:99-113. Crossref

7. Maslach C, Jackson SE. Maslach burnout inventory manual. 2nd ed. Mountain View, CA: Consulting Psychologists Press; 1986.

8. Jin MU, Jeong SH, Kim EK, Choi YH, Song KB. Burnout and its related factors in Korean dentists. Int Dent J 2015;65:22-31. Crossref

9. Gorter RC, Freeman R. Burnout and engagement in relation with job demands and resources among dental staff in Northern Ireland. Community Dent Oral Epidemiol 2011;39:87-95. Crossref

10. Denton DA, Newton JT, Bower EJ. Occupational burnout and work engagement: a national survey of dentists in the United Kingdom. Br Dent J 2008;205:E13;discussion 382-3.

11. Siu C, Yuen SK, Cheung A. Burnout among public doctors in Hong Kong: cross-sectional survey. Hong Kong Med J 2012;18:186-92.

12. Doolittle BR, Windish DM. Correlation of burnout syndrome with specific coping strategies, behaviors, and spiritual attitudes among interns at Yale University, New Haven, USA. J Educ Eval Health Prof 2015;12:41. Crossref

13. Jin WM, Zhang Y, Wang XP. Job burnout and organizational justice among medical interns in Shanghai, People’s Republic of China. Adv Med Educ Pract 2015;6:539-44.

14. Portero de la Cruz S, Vaquero Abellán M. Professional burnout, stress and job satisfaction of nursing staff at a university hospital. Rev Lat Am Enfermagem 2015;23:543-52. Crossref

15. Singh C, Cross W, Jackson D. Staff burnout—a comparative study of metropolitan and rural mental health nurses within Australia. Issues Ment Health Nurs 2015;36:528-37. Crossref

16. Wong MC, Ng KK, Tang AK, Tang GC, Lam VN. Occupational stress among dental practitioners in Hong Kong. Hong Kong Dent Assoc Millennium Rep 2001;II:50-5.

17. The Dental Council of Hong Kong. List of registered dentists under the General Register. Available from: http://www.dchk.org.hk/en/list/index.htm. Accessed 22 Oct 2015.

18. Cooper CL, Watts J, Kelly M. Job satisfaction, mental health, and job stressors among general dental practitioners in the UK. Br Dent J 1987;162:77-81. Crossref

19. Waddington TJ. New stressors for GDPs in the past 10 years. Br Dent J 1997;182:82-3. Crossref

20. Kopec JA, Esdaile JM. Occupational role performance in persons with back pain. Disabil Rehabil 1998;20:373-9. Crossref

21. Gorter RC, Eijkman MA, Hoogstraten J. Burnout and health among Dutch dentists. Eur J Oral Sci 2000;108:261-7. Crossref