Angiosarcoma is a rare form of sarcoma of vascular origin. It is notorious for its aggressive and relentless progression with frequent local recurrence and distant metastasis.1 Owing to its scarcity and innocuous appearance at an early stage mimicking an ordinary bruise or benign haemangioma, correct diagnosis is often delayed for several months. This problem is compounded further because most patients with angiosarcoma are elderly with frailty and co-morbidity, and prognosis after surgery as the definitive therapy is gloomy. The 5-year overall survival (OS) has been variously reported as 24% to 35%.1 2 When radiotherapy (RT) is given as the main treatment, median survival is only 8 months.3

Almost half (43%) of angiosarcomas originate from the skin of the head and neck.1 2 Compared with truncal and extremity angiosarcoma, the prognosis of cutaneous angiosarcoma (CAS) of the head and neck is even worse. Perez et al2 indicated a greater need for flap or graft reconstruction after tumour extirpation for head and neck CAS (HNCAS), a positive resection margin in 50%, and 5-year OS of 21.5%. Surgery was conventionally regarded as the mainstay of therapy for HNCAS. Because of the poor results and frequent margin involvement (78%),4 a multidisciplinary approach with adjuvant RT has been advocated.

To the best of our knowledge, most articles about HNCAS derived from a Caucasian population. We are not aware of any reported series from ethnic Chinese populations. We therefore conducted this retrospective study of HNCAS in the Chinese patients managed in our hospital over the past two decades. Demographic data, clinicopathological information, modality of treatment, and outcomes were reviewed. The latest approaches to the treatment of this devastating disease are also discussed.

Records of patients with histopathologically verified HNCAS from December 1997 to September 2016 were captured from the head and neck cancer registry of our department. Only Chinese patients were recruited. Patient data were prospectively collected and regularly updated in the registry.

Tumours were classified as a unifocal/localised versus multifocal/diffuse form. Unifocal/localised tumour was characterised by a discrete lesion without macroscopic satellitosis; it was considered operable and potentially curable. If gross satellite lesions were present or the main tumour was too extensive to be removed surgically, it was regarded as a multifocal/diffuse tumour that was probably incurable. Superficial tumours were those confined to the skin and subcutaneous tissue. Conversely, deep tumours were defined as transgression beyond the subcutaneous layer, for example encroaching on the underlying muscle or bone.

If the tumour was resectable, a wide excision with at least a 3-cm margin was performed. The defects were reconstructed by local scalp flap/ skin graft (scalp primary) or submental flap (face primary).5 Neck dissection was performed only in the presence of clinical or radiological evidence of nodal spread. Adjuvant RT was not routinely administered. When disease was deemed inoperable or the patient was unfit for surgery, palliative RT or chemotherapy would be considered.

The OS was calculated from the date of diagnosis to patient death or last follow-up, and is expressed in Kaplan-Meier curve. The data were computed using the SPSS (Windows version 20.0; IBM Corp, Armonk [NY], United States). The principles outlined in the 2013 version of the Declaration of Helsinki have been followed.

A total of 17 patients with HNCAS were identified and managed in our institution from December 1997 to September 2016. Their demographic and clinicopathological information are shown in Table 1. In brief, males predominated and there were only two female patients. Their median age was 81 years (range, 67-92 years). Only one patient had a history of whole-scalp RT for a lateral scalp angiosarcoma more than 10 years ago in another institution. He had a new angiosarcoma on the opposite side of his scalp that was treated in our centre as a second primary angiosarcoma induced by past RT. The median duration from onset of presentation to diagnosis was 4 months (range, 2-33 months).

Patients presented with protean symptoms: bleeding (n=7), pain (n=4), nodule (n=3), ulceration (n=2), pigmentation (n=2), pruritus (n=1), and localised oedema (n=1). One patient presented with an asymptomatic purplish macule but no other symptoms.

The tumours were present in the scalp only (n=11), face only (n=4), or both scalp and face (n=2). No patient had neck CAS. Ten patients had a localised/unifocal tumour. Seven patients were inflicted by multiple/diffuse lesions that were considered inoperable and treated with palliative intent in four. Of the three remaining patients with multiple/diffuse lesions, the tumours were still amenable to potentially curative surgery although two (cases 7 and 15) declined any treatment initially. Deep invasion occurred in two patients. Of the 11 patients whose tumour dimension had been documented, the median diameter was 4 cm (range, 3-13 cm). In the other six patients in whom dimensions were not recorded, there was extensive involvement by the HNCAS.

At the time of diagnosis, the numbers of T1 (≤5 cm) and T2 (>5 cm) diseases were seven and 10, respectively. Regional nodal spread was present in six patients. Only two patients (cases 2 and 3) were found to have distant metastases: lung in one patient, and lung and spine in the other (Table 1). The modalities of therapy were surgery (n=6), surgery + adjuvant RT (n=1), palliative RT (n=5), and palliative chemotherapy (n=3; two of them also received palliative RT following chemotherapy). The remaining two patients (cases 7 and 15) refused any form of treatment initially (Table 2). Incorporating subsequent therapy, surgery, RT and chemotherapy were eventually offered to seven, 10, and five patients, respectively.

Of the seven patients treated surgically, the resection margin was positive in two. Tumour recurred in six of them: four local and two regional recurrences. The median time to relapse was 7.5 months (range, 2-32 months). Overall, 16 patients had died; the causes of death were HNCAS in 12 and inter-current diseases in four (Table 2). One patient (case 15) was still living with the disease 21 months after diagnosis. The median OS was 11.1 months and the longest OS in our series was 42 months (case 6) [Fig 1].

The oncological outcome of this study is obviously far from satisfactory when compared with a reported median survival of 13.4 to 64 months as shown in Table 3,4 6 7 8 9 10 11 12 which summarises the postulated prognosticators. Only those studies performed over the past two decades and that focused on HNCAS with more than 10 cases were included. Patient age, tumour size, tumour differentiation, deep invasion, and margin status showed conflicting results. Of note, RT was the most promising and consistent prognosticator; only one study showed an adverse effect on survival.10 The disparity might well be due to selection bias for RT in a retrospective study—more advanced disease tends to receive adjuvant RT. The evidence lends credence to adjuvant RT for HNCAS. The suboptimal outcome in our study is multifactorial: detainment of diagnosis, inclusion of palliative cases for survival evaluation, advanced age precluding curative treatment, and unpopular adoption of RT or chemotherapy as multimodal therapy. In this series, adjuvant RT was administered to only one patient after surgery, either because others declined RT or wound complications precluded its application.

The diagnosis of HNCAS is often late as early lesions can simulate innocent violaceous macules masquerading as benign dermatological entities (Fig 2). From our experience, the median duration was 4 months prior to diagnostic confirmation with histopathological examination. Three of our patients (cases 7, 15, and 17) had their diagnosis made more than 12 months after onset of disease. The longest delay was 33 months (case 17) and is absolutely not acceptable. Interestingly, HNCAS manifested as pigmented lesions in two patients and thereby misled clinicians in decision making. Increased awareness of this rare disease by primary care clinicians is essential to expedite patient referral. For specialists, a low threshold to biopsy of newly developed purplish skin lesions in the elderly patients is pivotal to an early diagnosis.

In our series, patients who underwent palliative therapy were included in the OS calculation and this might have partially contributed to our poor results. This was echoed by Buschmann et al13 who also included patients with palliative resection in outcome evaluation; their longest survival reported was 36 months. This is similar to our experience where longest survival was 42 months. Conversely, Dettenborn et al10 reported 80 patients with HNCAS treated surgically (44 patients also received postoperative RT) with curative intent and 5-year OS of 54% and median OS of 64 months. Similarly, Suzuki et al12 described their experience of definitive RT as the principal curative treatment for HNCAS; a median OS survival of 31 months was attained. Nonetheless, the results of RT were compromised when palliative cases were incorporated: only 12% of patients survived more than 5 years in one study.14

Our patients were older (median age, 81 years) than those reported in the literature (median age, 63-77 years) [Table 3]. The prognostic significance of age on the outcome of HNCAS is controversial. Patel et al11 reported that patients younger than 70 years fared better with improved locoregional control and relapse-free survival than older patients. Although age is not a confirmed prognostic factor, advanced age often precludes patients from curative therapy due to concomitant chronic diseases or general debility. The elderly patients are also prone to dying of other disease as encountered in this series; four patients died of inter-current illnesses (Table 2). Nonetheless, effective systemic treatments can be well tolerated by some elderly patients with HNCAS. With taxane-based regimens, a response rate of 83% (10 out of 12 patients) was achieved and the progression-free survival was approximately 7 months.15

Surgery is historically the main treatment for HNCAS. The latest approach to optimal management of HNCAS is combined treatment encompassing surgery, RT, or chemotherapy. Adjuvant RT should be liberally offered to maximise the oncological outcomes following surgery. As shown in Table 3, five of six studies support the beneficial role of RT.4 6 7 8 10 11 Guadagnolo et al7 advocated simple resection of the gross tumour to facilitate non-complicated reconstruction and thus expedite RT. Two-staged surgery was discouraged. The resection margin status was not critical to survival if timely adjuvant RT was administered. From their experience, the 5-year OS was 43% and 5-year disease-specific survival was 46%. In a review article, Hwang et al16 also concluded that positive margin was common (64%) but did not impair the ultimate outcome. They recommended that surgeons should not be too obsessive about removing each and every cancer cell if RT was to be pursued.

Definitive treatment with RT and/or chemotherapy has also been reported to be effective for HNCAS.9 Of 17 patients treated in that study, complete and partial responses were accomplished in none and five patients, respectively. The median OS was 26 months. Multimodality treatment in various combinations with surgery, RT, and chemotherapy has been asserted by Patel et al11 to be effective in improving locoregional control, relapse-free survival, and OS. In another study, survival (37 months) following combined therapy (RT and chemotherapy) was better than either modality alone: 23 months for RT and 15 months for chemotherapy.16 Docetaxel is the preferred agent due to its antiangiogenic and radio-sensitising effects.9 Other agents have also been successfully used and include doxorubicin, ifosfamide, bevacizumab, and interleukin-2. Systemic treatment may be used in a neoadjuvant setting, adjuvant setting, and as concurrent treatment with RT.8 17 18 Conversely, RT plus chemotherapy was not shown to have any prognostic value in a meta-analysis by Shin et al.19 Further studies should be carried out to elucidate the benefit of combined modality treatment.

Ip and Lee20 reported a smaller local series of CAS that was not confined to the head and neck. A total of seven patients were enrolled from three clinics of the Social Hygiene Service in Hong Kong. Only six patients had HNCAS. Similarly, poor prognosis was demonstrated in these patients. Our bigger series focusing on HNCAS provides a more updated and detailed strategy for the management of this rare disease in ethnic Chinese.

Our study has some limitations. First, the sample size was small as only 17 patients were included for evaluation. Nevertheless it is the largest series reported in our locality. The study spanned over 19 years (December 1997 to September 2016) and treatment has evolved over this period. Moreover, like many retrospective series, recall bias or selection bias are inherent limitations. Patient symptoms, signs, presentation duration, and criteria for treatment might not be completely accurate.

We present the first report of HNCAS in ethnic Chinese. The oncological outcome is far from satisfactory. A high index of suspicion is mandatory for prompt diagnosis of early disease. Adjuvant RT, as supported by evidence from the literature, is recommended following surgery that should aim at gross tumour extirpation to ensure uneventful reconstruction as well as timely implementation of RT. The benefit and role of systemic treatment in various combinations with surgery or RT require further study.

All authors have disclosed no conflicts of interest.

1. Mark RJ, Poen JC, Tran LM, Fu YS, Juillard GF. Angiosarcoma. A report of 67 patients and a review of the literature. Cancer 1996;77:2400-6. Crossref

2. Perez MC, Padhya TA, Messina JL, et al. Cutaneous angiosarcoma: a single-institution experience. Ann Surg Oncol 2013;20:3391-7. Crossref

3. Sasaki R, Soejima T, Kishi K, et al. Angiosarcoma treated with radiotherapy: impact of tumor type and size on outcome. Int J Radiat Oncol Biol Phys 2002;52:1032-40. Crossref

4. Pawlik TM, Paulino AF, Mcgini CJ, et al. Cutaneous angiosarcoma of the scalp: a multidisciplinary approach. Cancer 2003;98:1716-26. Crossref

5. Chow TL, Chan TT, Chow TK, Fung SC, Lam SH. Reconstruction with submental flap for aggressive orofacial cancer. Plast Reconstr Surg 2007;120:431-6. Crossref

6. Aust MR, Olsen KD, Lewis JE, et al. Angiosarcoma of the head and neck: clinical and pathologic characteristics. Ann Otol Rhinol Laryngol 1997;106:943-51. Crossref

7. Guadagnolo BA, Zagars GK, Araujo D, Ravi V, Shellenberger TD, Sturgis EM. Outcomes after definitive treatment for cutaneous angiosarcoma of the face and scalp. Head Neck 2011;33:661-7. Crossref

8. Ogawa K, Takahashi K, Asato Y, et al. Treatment and prognosis of angiosarcoma of the scalp and face: a retrospective analysis of 48 patients. Br J Radiol 2012;85:e1127-33. Crossref

9. Miki Y, Tada T, Kamo R, et al. Single institutional experience of the treatment of angiosarcoma of the face and scalp. Br J Radiol 2013;86:20130439. Crossref

10. Dettenborn T, Wermker K, Schulze HJ, Klein M, Schwipper V, Hallermann C. Prognostic features in angiosarcoma of the head and neck: a retrospective monocenter study. J Craniomaxillofac Surg 2014;42:1623-8. Crossref

11. Patel SH, Hayden RE, Hinni ML, et al. Angiosarcoma of the scalp and face: the Mayo Clinic experience. JAMA Otolaryngol Head Neck Surg 2015;141:335-40. Crossref

12. Suzuki G, Yamazaki H, Takenaka H, et al. Definitive radiation therapy for angiosarcoma of the face and scalp. In Vivo 2016;30:921-6.Crossref

13. Buschmann A, Lehnhardt M, Toman N, Preiler P, Salakdeh MS, Muehlberger T. Surgical treatment of angiosarcoma of the scalp: less is more. Ann Plast Surg 2008;61:399-403. Crossref

14. Holden CA, Spittle MF, Jones EW. Angiosarcoma of the face and scalp, prognosis and treatment. Cancer 1987;59:1046-57. Crossref

15. Letsa I, Benson C, Al-Muderis O, Judson I. Angiosarcoma of the face and scalp: effective systemic treatment in the older population. J Geriatr Oncol 2014;5:276-80. Crossref

16. Hwang K, Kim MY, Lee SH. Recommendations for therapeutic decisions of angiosarcoma of the scalp and face. J Craniofac Surg 2015;26:e253-6. Crossref

17. Wollina U, Fuller J, Graefe T, Kaatz M, Lopatta E. Angiosarcoma of the scalp: treatment with liposomal doxorubicin and radiotherapy. J Cancer Res Clin Oncol 2001;127:396-9. Crossref

18. Yang P, Zhu Q, Jiang F. Combination therapy for scalp angiosarcoma using bevacizumab and chemotherapy: a case report and review literature. Chin J Cancer Res 2013;25:358-61.

19. Shin JY, Roh SG, Lee NH, Yang KM. Predisposing factors for poor prognosis of angiosarcoma of the scalp and face: systemic review and meta-analysis. Head neck 2017;39:380-6. Crossref

20. Ip FC, Lee CK. Cutaneous angiosarcoma: a case series in Hong Kong. Hong Kong J Dermatol Venereol 2010;18:6-14.

Chronic kidney disease (CKD) is the progressive loss of kidney function over a period of time. End-stage renal disease (ESRD) is the final stage of CKD. Patients with ESRD require renal replacement therapy that includes haemodialysis, peritoneal dialysis, and renal transplantation.

Currently, there are approximately 7000 patients on various forms of renal replacement therapy being cared for in the public sector in Hong Kong. As of 31 December 2016, 2047 patients were on the renal transplant waiting list. Nonetheless, between 2007 and 2016, only 58 to 87 cadaveric renal transplants were performed in Hong Kong each year.1 With the long waiting list and low number of cadaveric kidneys available, living donor renal transplant is the only possible alternative. It offers advantages over other renal replacement therapies, as it provides better long-term results, shortens the waiting time for an organ, lowers the risk of complications or rejection, and provides better quality of life after recovery. Despite these advantages, only seven to 15 living donor transplants were performed each year between 2007 and 2015 at the hospitals of the Hong Kong Hospital Authority.1

One of the major fears of an individual who is considering living organ donation concerns possible clinical outcomes. Although studies show that living donors have a similar to or better life expectancy than the general population, they are nevertheless at increased risk of developing ESRD, hypertension, gestational hypertension, and pre-eclampsia.2 3 4

In Hong Kong, few reports on the perioperative, short-term, and long-term clinical outcomes are available, especially those related to the minimally invasive surgical approach now employed for donor nephrectomy. This study reports our observation of characteristics of donors, and the short- and long-term clinical outcomes following living donor nephrectomy in Hong Kong.

We retrospectively reviewed the characteristics and short- and long-term clinical outcomes of all patients who underwent living donor nephrectomy at the Prince of Wales Hospital in Hong Kong between January 1990 and December 2015. Information was obtained from the Clinical Management System that includes the majority of electronic patient records—including consultation histories, operation records, radiology results, laboratory results, and medication records—collected and filed under the Hospital Authority since 2000. Medical records before 2000 and pregnancy-related information were reviewed manually by formally trained medical students and cross-checked by a urologist, and retrieved from the medical records of the involved patients.

The study was conducted in accordance with the principles outlined in the Declaration of Helsinki, and approved by the Joint Chinese University of Hong Kong–New Territories East Cluster Clinical Research Ethics Committee, with the requirement of patient informed consent waived because of its retrospective nature.

Baseline demographics including sex, age at donation, ethnicity, relationship with recipient, diabetes mellitus status, hypertension status, body mass index, and serum creatinine level were obtained. Glomerular filtration rate (GFR) was derived from the serum creatinine level using a modified equation from the Modification of Diet in Renal Disease (MDRD) study.5 Operation details, including surgical approach, laterality of donated kidney, operating time, warm ischaemia time, blood loss, and need for transfusion were retrieved.

Short-term complications within 30 days of surgery were classified according to the Clavien-Dindo classification of surgical complications.6 Long-term outcomes were also assessed, with particular reference to development of hypertension, diabetes mellitus, renal stones, proteinuria, and renal failure. Serial changes in GFR were also assessed.

For female donors, pregnancy-related variables were recorded and included any pregnancy after surgery, records of pregnancy-related hydronephrosis, pregnancy-related urinary tract infection, pre-eclampsia, gestational diabetes mellitus, gestational hypertension, and any fetal loss.

All statistical analyses were performed using the SPSS (Windows version 23.0; IBM Corp, Armonk [NY], United States). Categorical variables were presented in counts and percentages while continuous variables were presented as mean ± standard deviation. Outcomes following open and laparoscopic techniques were compared by Chi squared test or Fisher’s exact test for categorical variables, and independent t test or Mann-Whitney U test for continuous variables. Paired t test or Wilcoxon rank sum test, whichever was appropriate, was used to evaluate the pre- and post-difference in GFR. A P value of <0.05 was considered statistically significant. Missing data were excluded from analysis.

Between 1 January 1990 and 31 December 2015, a total of 83 donors underwent unilateral nephrectomy at the Prince of Wales Hospital. In one donor, records could not be traced, with only information about the sex, age at nephrectomy, and type of surgical technique.

Of the 83 donors, 56 (67.5%) were female. The mean age at nephrectomy was 37.3 ± 10.0 years. The majority were Chinese (97.6%) and a first-degree relative of the recipient (79.3%). None had hypertension or diabetes mellitus. The mean preoperative GFR was 96.0 ± 17.5 mL/min/1.73 m². Nine (11.0%) donors had thalassaemia trait, four (4.9%) had hepatitis B, and two (2.4%) had asthma (Table 1).

Around half (n=44, 53.0%) of the donors underwent open living donor nephrectomy, as this was the only technique used at our centre until 2002. After 2002, a hand-port assisted laparoscopic approach (n=4, 4.8%) and later a laparoscopic approach (n=35, 42.2%) were adopted. In most instances, the left kidney was donated (n=77, 93.9%) [Table 2].

Comparing laparoscopic or hand-port assisted laparoscopic living donor nephrectomy (LDN) with open donor nephrectomy (ODN), LDN was associated significantly with longer warm ischaemia time (309.0 ± 113.0 s vs 62.0 ± 17.9 s; P<0.01), less blood loss (55.3 ± 33.7 mL vs 532.2 ± 270.0 mL; P<0.01), and shorter hospital stay (5.7 ± 2.0 days vs 8.1 ± 1.9 days; P<0.01). In addition, LDN was associated significantly with more short-term complications (53.8% vs 20.9%; P<0.01). The most commonly experienced complication was epigastric pain/nausea and vomiting (n=18, 56.3%), followed by fever requiring medication (n=4, 12.5%). Most complications were grade 1 on the Clavien-Dindo classification scale (n=16, 50.0%), only three (9.4%) were grade 3a and all were related to ODN. The grade 3a complications were wound dehiscence that required a second operation for re-suturing, persistent pancreatic fluid discharge that required insertion of a pancreatic stent, and pneumothorax with chest drain inserted.

The mean follow-up time was 12.0 ± 8.3 years. The mean GFR was 96.0 ± 17.5 mL/min/1.73 m² at baseline and it dropped significantly to 66.8 ± 13.5 mL/min/1.73 m² at 1-year follow-up (P<0.01). The GFR then gradually improved until 8 years after surgery and became stable (Fig). Of 73 living donors with at least one follow-up (mean follow-up time, 12.0 ± 8.2 years) and baseline serum creatinine level available, the latest GFR was 75.6% ± 15.1% of baseline GFR with the mean latest GFR being 71.3 ± 14.2 mL/min/1.73 m². The mean GFR was 70.4% ± 12.3% of baseline level 1 year after surgery. Comparison of latest GFR with that 1 year after surgery revealed that it was stable (± 10% change) in 23 (39.0%) of 59 patients and higher (>10% increment) in 29 (49.2%) patients. None of the donors had died or developed ESRD. Fourteen (18.2%) donors developed hypertension, two (2.6%) had diabetes mellitus, and three (4.0%) had experienced proteinuria (Table 3).

Of 56 female donors, 11 (19.6%) became pregnant after kidney donation: 17 pregnancies were reported. None of the pregnant donors experienced gestational hydronephrosis or gestational hypertension. Three donors each had gestational diabetes mellitus, pre-eclampsia, and post-delivery urinary tract infection. Two donors had experienced fetal loss, one in the first trimester and another one at an unknown gestational age (Table 4).

Postoperative morbidity and mortality are the prime concerns when making a decision about kidney donation. Our results confirm that living donor nephrectomy is a relatively safe procedure, with a low incidence of major complications and mortality. In addition, the incidence of developing any other major disease was not particularly high in our series. This form of renal replacement therapy should be further promoted in Hong Kong to benefit more people with ESRD.

Results from previous studies have shown that living renal donors have a similar to or better life expectancy than the general population.7 8 9 10 Mjøen et al,11 however, reported that compared with healthy matched individuals, living renal donors had an increased risk of death. In Hong Kong, Chu et al12 reported one death related to multiple myeloma among 95 living renal donors with active follow-up and a mean follow-up period of 13.4 years. There were no deaths recorded in our study with a mean follow-up of 12 years.

Long-term renal function is another major concern of renal donors. Our results revealed that 1 year after living donor nephrectomy, the mean GFR of the kidney donors dropped significantly from 96.0 ± 17.5 mL/min/1.73 m² at baseline to 66.8 ± 13.5 mL/min/1.73 m². Nonetheless, it then gradually improved. This is probably partly related to the adaptation of the remaining kidney with hyper-filtration. From our series, the mean GFR was 70.4% ± 12.3% of baseline level 1 year after surgery but improved to 75.6% ± 15.1% of baseline level at the last follow-up. In the majority (88.2%) of donors, the last available GFR was static or higher than that 1 year after donation. This is comparable with the report of Rook et al13 in which GFR usually reached 64% ± 7% of the pre-donation level 1 year after donation.

Despite these changes in GFR, ESRD in renal donors is very rare, with an incidence of less than 0.5% in 15 years after donation.11 14 15 Ibrahim et al8 reported that survival and risk of ESRD in kidney donors appeared to be similar to those in the general population. Our study and that of Chu et al12 observed no ESRD in local kidney donors.

The effect of kidney donation on the development of hypertension is controversial. Although reports suggest that the incidence of hypertension among kidney donors increases,16 17 18 19 others have not confirmed this observation.20 21 22 23 24 In Hong Kong, the prevalence of hypertension in the general population was 12.6% in 2014,25 which is lower than our reported figure of 18.2%. With the progression of time after surgery, however, the prevalence of hypertension among living donors is expected to increase as age is a known influence in hypertension. Without a comparable control group, we cannot conclude if there is any actual discrepancy in the prevalence of hypertension among living donors compared with the general population.

Young female potential donors may have concerns about the impact of kidney donation on any future pregnancy. Garg et al4 reported that gestational hypertension or pre-eclampsia was more common among living donors than non-donors. Although our study showed an alarmingly high percentage (11%) of pre-eclampsia and absence of gestational hypertension, the small sample size (11 donors reported one or more pregnancies) undermines the ability to infer the actual percentage.

Perioperative complications may also deter potential living donors. Based on the US data, Lentine et al26 reported that 16.8% of donors experience a perioperative complication; most commonly gastrointestinal (4.4%). Our study showed a higher complication rate of 36.6%, with epigastric pain or nausea and vomiting being the major complication (56.3%). We further examined the techniques used and established that the complication rates of 20.9% or 53.8% respectively in donors who underwent ODN or LDN were significantly different (P<0.01). Despite the above mean complication rates, all complications of LDN were mild and of grade 1 or 2 according to the Clavien-Dindo classification, while three patients who underwent ODN had grade 3a complications. This is contrary to the majority of previous findings that suggest a lower perioperative complication rate for LDN and increased risk of more serious complications than during an ODN,27 although other indicators such as longer warm ischaemia time, less blood loss, and shorter hospital stays were still in line with previous findings. Further analysis of the differences between our local data and those of previous studies is warranted.

This study has several limitations. First, this was a retrospective study and the total number of living donors was restricted. Second, data quality could not be controlled and some data were incomplete, in particular for the obstetric records at other hospitals. Some data were also lost either because records were too old and pre-dated the electronic system or donors were no longer followed up at our centre. The oldest record included in the study was from 1990. At that time, record keeping was not always accurate, resulting in some baseline records from the early 1990s being missing. For example, the baseline GFR level of seven (8.4%) patients was not found, and might have affected the overall data quality as well as the analysis and conclusion. Third, although the urologist endeavoured to ensure accurate data entry, initial interpretation of the raw records was by medical students so certain inaccuracies might have occurred. Lastly, it is known that GFR might be underestimated when derived from the MDRD equation.

Living donor kidney transplantation is an important approach to improve the quality of life of patients with ESRD. Good short- and long-term outcome for kidney donors is important for promoting kidney donation. Our results suggest that the overall perioperative outcomes are good, with only very few serious (grade III) complications after surgery, occurring following an open approach. Long-term kidney function of donors was satisfactory and no patients developed ESRD. Although we had no control arm in our study, the overall incidences of new-onset medical diseases and pregnancy-related complications were low. The introduction of a laparoscopic approach for kidney harvesting has helped to decrease blood loss during surgery and also shorten hospital stay. Based on this encouraging result, relatives of patients with ESRD should be encouraged to consider the possibility of kidney donation.

Sincere thanks are given to Ms Karen Man-ting Chuk, Ms Tracy Lok-sze Chiu, Mr Wing-tung Leung, and Mr On-wa Ng for assisting with the data collection.

All authors have disclosed no conflicts of interest.

1. Statistics (Milestones of Hong Kong organ transplantation): organ donation. Available from: http://www.organdonation. gov.hk/eng/statistics.html. Accessed 27 Dec 2017.

2. Reese PP, Boudville N, Garg AX. Living kidney donation: outcomes, ethics, and uncertainty. Lancet 2015;385:2003-13. Crossref

3. Delanaye P, Weekers L, Dubois BE, et al. Outcome of the living kidney donor. Nephrol Dial Transplant 2012;27:41- 50. Crossref

4. Garg AX, Nevis IF, Mcarthur E, et al. Gestational hypertension and preeclampsia in living kidney donors. N Engl J Med 2015;372:124-33. Crossref

5. Levey AS, Greene T, Kusek JW, Beck GJ. A simplified equation to predict glomerular filtration rate from serum creatinine. J Am Soc Nephrol 2000;11:155A0828.

6. Dindo D, Demartines N, Clavien PA. Classification of surgical complications: a new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann Surg 2004;240:205-13. Crossref

7. Segev DL, Muzaale AD, Caffo BS, et al. Perioperative mortality and long-term survival following live kidney donation. JAMA 2010;303:959-66. Crossref

8. Ibrahim HN, Foley R, Tan L, et al. Long-term consequences of kidney donation. N Engl J Med 2009;360:459-69. Crossref

9. Okamoto M, Akioka K, Nobori S, et al. Short- and long-term donor outcomes after kidney donation: analysis of 601 cases over a 35-year period at Japanese single center. Transplantation 2009;87:419-23. Crossref

10. Garg AX, Meirambayeva A, Huang A, et al. Cardiovascular disease in kidney donors: matched cohort study. BMJ 2012;344:e1203. Crossref

11. Mjøen G, Hallan S, Hartmann A, et al. Long-term risks for kidney donors. Kidney Int 2014;86:162-7. Crossref

12. Chu KH, Poon CK, Lam CM, et al. Long-term outcomes of living kidney donors: a single centre experience of 29 years. Nephrology (Carlton) 2012;17:85-8. Crossref

13. Rook M, Hofker HS, van Son WJ, Homan van der Heide JJ, Ploeg RJ, Navis GJ. Predictive capacity of pre-donation GFR and renal reserve capacity for donor renal function after living kidney donation. Am J Transplant 2006;6:1653-9. Crossref

14. Muzaale AD, Massie AB, Wainwright J, McBride MA, Wang M, Segev DL. Long-term risk of ESRD attributable to live kidney donation: matching with healthy non-donors. Am J Transplant 2013;13:204-5.

15. Fehrman-Ekholm I, Nordén G, Lennerling A, et al. Incidence of end-stage renal disease among live kidney donors. Transplantation 2006;82:1646-8. Crossref

16. Kasiske BL, Ma JZ, Louis TA, Swan SK. Long-term effects of reduced renal mass in humans. Kidney Int 1995;48:814-9. Crossref

17. Gossmann J, Wilhelm A, Kachel HG, et al. Long-term consequences of live kidney donation follow-up in 93% of living kidney donors in a single transplant center. Am J Transplant 2005;5:2417-24. Crossref

18. Garg AX, Prasad GV, Thiessen-Philbrook HR, et al. Cardiovascular disease and hypertension risk in living kidney donors: an analysis of health administrative data in Ontario, Canada. Transplantation 2008;86:399-406. Crossref

19. Doshi MD, Goggins MO, Li L, Garg AX. Medical outcomes in African American live kidney donors: a matched cohort study. Am J Transplant 2012;13:111-8. Crossref

20. Fehrman-Ekholm I, Dunér F, Brink B, Tydén G, Elinder CG. No evidence of accelerated loss of kidney function in living kidney donors: results from a cross-sectional follow-up. Transplantation 2001;72:444-9. Crossref

21. Macdonald D, Kukla AK, Ake S, et al. Medical outcomes of adolescent live kidney donors. Pediatric Transplant 2014;18:336-41. Crossref

22. Janki S, Klop KW, Dooper IM, Weimar W, Ijzermans JN, Kok NF. More than a decade after live donor nephrectomy: a prospective cohort study. Transpl Int 2015;28:1268-75. Crossref

23. Tavakol MM, Vincenti FG, Assadi H, et al. Long-term renal function and cardiovascular disease risk in obese kidney donors. Clin J Am Soc Nephrol 2009;4:1230-8. Crossref

24. El-Agroudy AE, Wafa EW, Sabry AA, et al. The health of elderly living kidney donors after donation. Ann Transplant 2009;14:13-9.

25. Census and Statistics Department, Hong Kong SAR Government. Thematic Household Survey Report No. 58. Available from: http://www.statistics.gov.hk/pub/ B11302582015XXXXB0100.pdf. Accessed 28 Oct 2016.

26. Lentine KL, Lam NN, Axelrod D, et al. Perioperative complications after living kidney donation: a national study. Am J Transplant 2016;16:1848-57. Crossref

27. Fonouni H, Mehrabi A, Golriz M, et al. Comparison of the laparoscopic versus open live donor nephrectomy: an overview of surgical complications and outcome. Langenbecks Arch Surg 2014;399:543-51. Crossref

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