Gonadectomy in conditions affecting sex development: a registry-based cohort study

Objectives: To determine trends in clinical practice for individuals with DSD requiring gonadectomy. Design: Retrospective cohort study. Methods: Information regarding age at gonadectomy according to diagnosis; reported sex; time of presentation to specialist centre; and location of centre from cases reported to the International DSD Registry and who were over 16 years old in January 2019. Results: Data regarding gonadectomy were available in 668 (88%) individuals from 44 centres. Of these, 248 (37%) (median age (range) 24 (17, 75) years) were male and 420 (63%) (median age (range) 26 (16, 86) years) were female. Gonadectomy was reported from 36 centres in 351/668 cases (53%). Females were more likely to undergo gonadectomy ( n = 311, P < 0.0001). The indication for gonadectomy was reported in 268 (76%). The most common indication was mitigation of tumour risk in 172 (64%). Variations in the practice of gonadectomy were observed; of the 351 cases from 36 centres, 17 (5%) at 9 centres had undergone gonadectomy before their first presentation to the specialist centre. Median age at gonadectomy of cases from high-income countries and low-/middle-income countries (LMIC) was 13.0 years (0.1, 68) years and 16.5 years (1, 28), respectively ( P < 0.0001) with the likelihood of long-term retention of gonads being higher in LMIC countries. Conclusions: The likelihood of gonadectomy depends on the underlying diagnosis, sex of rearing and the geographical setting. Clinical benchmarks, which can be studied across all forms of DSD will allow a better understanding of the variation in the practice of gonadectomy.


Introduction
Differences or disorders of sex development (DSD) are a heterogeneous group of conditions, that can be associated with gonadal dysfunction and germ cell cancers (1). The estimated prevalence of germ cell tumours ranges from 0.8 to 40% depending on age and underlying diagnosis (2,3). To mitigate this tumour risk, as well as for avoiding the possibility of hormone production that is discordant to the gender, children and adolescents with DSD have often undergone gonadectomy (4,5). However, when counselling patients as well as their parents in case of a child with DSD, a careful balance must be sought, between mitigating this risk, preserving any residual gonadal function and encouraging patient participation in decisions regarding gonadectomy (4). Whilst single centre and single condition reports of the experience of gonadectomy in cases of DSD do exist (5,6,7,8), there is little information on variations in care as well as any emerging trends in the timing of gonadectomy. More recently, there has been an increasing emphasis on delaying gonadectomy in a range of conditions associated with DSD (9) especially when there is a possibility that informed consent cannot be obtained from the young patient (10). The primary aim of the current study was to use the I-DSD Registry to investigate recent trends in gonadectomy internationally for a wider range of conditions associated with DSD. In addition, the use of objective outcomes including age at gonadectomy, sex of rearing and time of presentation was explored.

Methods
The I-DSD Registry was interrogated for information regarding year of birth, age at presentation to a specialist centre, the diagnosis, karyotype and sex of rearing of all cases who were over the age of 16.0 years in January 2019 and who had a disorder of androgen action, androgen synthesis or gonadal development; Leydig cell hypoplasia; persistent Müllerian duct syndrome or a non-specific DSD. The I-DSD registry is an international database of pseudo-anonymised information on patients with DSD and is approved by the National Research Ethics Service in the United Kingdom (IRAS 269776 and Research Ethics Committee reference 19/WS/0131) as a research database of information that is collected as part of routine clinical care at specialist DSD centres (11) and further details are available at www.i-dsd.org. Throughout the manuscript, the terms male/female refer to reported sex at the time of the study, as reported by clinicians on the I-DSD Registry. The cut-off of 16 years was chosen to enable a reasonable time-lapse between diagnosis and gonadectomy to better understand when gonadectomies were performed. For eligible cases, the reporting clinicians were approached for information on gonadectomy, and in those cases that had undergone gonadectomy, additional data were collected on age at gonadectomy, indication for gonadectomy and histology result, if available.

Data management and statistical analysis
The data received from each centre were collated and stored on Microsoft Excel version 16.4 and accessed only by ALH and SFA, as the principal investigators who planned the original data analysis. Where incomplete data were available, for example lack of information regarding indication or histology, this was indicated. Data were quality checked to ensure no duplicate patients were entered, for example in the case of a patient moving care from one I-DSD centre to another. To assess temporal differences, comparisons were performed between individuals born before 1999 and after 1999, to approximate with the increasing awareness of the John vs Joan case (12,13). For assessment of geographical differences in practice, participating centres were categorised as those from a low-or middle-income country (LMIC) or from a high-income country (HIC) as defined by the 2019 World Bank classification (14). Continuous variables were described as median and ranges and normality was assessed via the Shapiro-Wilk test. Inter-group comparison for these variables was performed by the Mann-Whitney U test or one way ANOVA as appropriate. The Fisher exact test was performed to compare proportions in different groups. Linear regression was performed to assess associations between trends of gonadectomy and odds ratios calculated for likelihood of gonadectomy. The level of P < 0.05 was considered to be statistically significant and all analyses were performed using Graphpad Prism 8 (San Diego, CA, 2018).

Age at gonadectomy
Of the 351 cases who had gonadectomy, data on both, age at gonadectomy and age at reported time of presentation, was available in 296 (79%) ( Table 1). As expected, age at gonadectomy was associated with the age at presentation (P < 0.0001) with the age at gonadectomy increasing with an increasing age at presentation, irrespective of the underlying diagnosis (Fig. 1A). However, the underlying diagnosis when coupled with the sex of rearing, did influence the interval between age at presentation and age at gonadectomy such that females with CAIS were more likely to be older at gonadectomy compared to age at presentation (Fig. 1A). Overall, females had gonadectomy later, at a median age of 15 years (0.1, 68), compared to males who had gonadectomy at a median age of 9 years (0.1, 54) (P = 0.047) ( Table 1). There were no statistically significant differences between diagnoses. There were no statistically significant trends for a change in the age at gonadectomy according to indication for gonadectomy ( Fig. 1B) or over time (Fig. 1C). Of those who were diagnosed under the age of 10 yrs, there was no difference in age at gonadectomy in those who were born prior to 1990 compared to those who were born between 1990 and 1999 (median age 5 years vs 6 years, P = 0.2).

Gonadectomy prior to presentation to specialist centre
Of the 351 cases of gonadectomy where information on age at gonadectomy was available, 17 (5%) (16 females and 1 male), had gonadectomy performed before presentation to the specialist centre ( Fig. 2A).
Of these 17 patients, 6 (38%) had CAIS; 4 (24%) had CGD; 3 (18%) had PAIS and there was one case each of 17β-dehydrogenase type 3 deficiency, 5-α-reductase type 2 deficiency, Leydig cell hypoplasia and non-specific XY DSD. The median reported age at presentation for these individuals was 16 years (3, 38) with a median period of 4 years (0.2, 23) from gonadectomy to presentation to specialist centre. Only 6 of these 17 (35%) cases were born after 1990 but linear regression analysis did not identify Age at gonadectomy according to diagnosis, year of birth and age at presentation to specialist centre. a statistically significant temporal trend for a change in the number of individuals with gonadectomy prior to presentation according to the year of birth (P = 0.3) (Fig. 2B). The 17 cases that had gonadectomy prior to reported presentation to the specialist centre were from 9/36 (34%) centres (Fig. 2C). The proportion of registered patients from each centre who underwent gonadectomy prior to presentation to the specialist centre varied from 0 to 100% (Fig. 2D). There was a wide range of variability in the number of gonadectomies performed per centre, with the median being 1 (1, 126) although there was no statistically significant difference in the number of cases with gonadectomy before specialist presentation in those centres that reported more cases of gonadectomy (Fig. 2E).
The median age at diagnosis of cases from HIC and LMIC was 6 years (0.1, 53) and 14 years (0.1, 38), respectively (P = 0.015) and the median age at gonadectomy of cases from HIC and LMIC was 13 years (0.1, 68) years and 16.5 years (1, 28), respectively (P < 0.0001). Of the 95 cases in LMIC centres, a disorder of gonadal development was the commonest condition in 43 cases (45%). In HIC, the commonest group of conditions was a disorder of androgen action in 131 cases (51%) ( (1.5-11.9, P < 0.01). There were no significant differences in age at presentation or age at gonadectomy according to underlying diagnosis in LMIC centres vs HC centres. There was no statistically significant difference in the number of cases with gonadectomy prior to presentation to the specialist centre in those from LMIC centres vs HIC centres. Of the 17 cases with gonadectomy prior to specialist presentation, 5 (29%) were from 4 LMIC centres and 12 from 5 HIC centres.
1 case of partial gonadal dysgenesis) and dysgerminoma in 2 (18%) (1 case each of ovotesticular DSD and CGD). Data on histology were available for 9 out of 17 cases (53%) who had gonadectomy prior to presentation to the specialist centre and none of these had any histological evidence of tumourigenesis.

Discussion
This study demonstrates differences that exist in the practice of gonadectomy and that are dependent on the underlying diagnosis, the sex of rearing and the age at presentation. Clear patterns have previously been reported for conditions such as CAIS, where high rates of affected women have undergone gonadectomy (15) and these were reinforced in the current study. However, previous reports of gonadectomy in other forms of DSD, for example 17-β dehydrogenase type 3 deficiency, 5-α reductase type 2 deficiency or non-specific XY DSD are rare. In some of these conditions that are associated with a disorder of androgen synthesis, the indication for early gonadectomy was the avoidance of pubertal virilisation in an XY girl. Given that more children with XY DSD are being raised as boys (12), there is a greater prospect of preserving gonadal function and fertility (16,17) and there is a greater emphasis on the patient's own views on gonadectomy (4,18,19), there has been an increased emphasis on retaining gonads until the age when patients with DSD can decide for themselves (20). Mitigation of tumour risk was the commonest reported indication for performing gonadectomy and therefore as expected, in the current study, the number of cases that actually had any evidence of tumour development was very small. It is also noteworthy, that the conditions that were associated with a tumour included a case of PMDS and ovotesticular DSD, two conditions where the risk of gonadal tumourigenesis has generally been considered to be low (21,22). Thus, the study emphasizes the need for ongoing monitoring of these outcomes in those conditions that are traditionally thought to be low risk. The study also demonstrated that a substantial proportion of cases continued to retain gonads into adulthood, even in conditions such as gonadal dysgenesis, where the risk of germ cell tumours is particularly high (18,23).
With the shifts in societal attitudes, it would be useful to prospectively monitor the age at presentation and the age at gonadectomy. The current study showed that, in about 1 in 20 cases, the age at gonadectomy was prior to the age at presentation to a specialist DSD centre. There may be many reasons why this was the case. It is assumed that all of these individuals had specialists involved in their care outwith the DSD reference centre but a prospective study would allow this to be monitored and it is hoped that with improved awareness of DSD conditions, fewer affected individuals are being referred late to specialists in this field. It was also noteworthy that there were variations in practice between LMIC and HIC centres with the proportion of cases that had a gonadectomy being lower in LMIC countries. The older age at gonadectomy at the LMIC centres may be secondary to an older age at presentation in these cases but the lower proportion of cases undergoing gonadectomy may be due to difficulties in access to specialist diagnostic or surgical services (19,24) or due to differences in cultural and societal attitudes to gonadectomy (25). These aspects as well as the surveillance measures that are used for monitoring retained gonads will need to be explored in future studies.
Over the last few decades, it has become increasingly clear that the risk of tumour development in conditions associated with DSD is dependent on the underlying condition. It is likely that the decision to perform gonadectomy is influenced by this knowledge as well as other variables, as above. Thus, the management of gonads requires a detailed understanding of all these factors which can only be provided by a specialist team that can call on this multidisciplinary expertise (4). The I-DSD Registry represents a valuable resource that can facilitate the study of clinical practice in this large heterogenous group of rare conditions. Studies using this resource have previously revealed the global shift in the practice of sex assignment (12) and more recent studies have shown how practices vary in the management of the testes in people with AIS (15). The current study has taken this further by showing how these differences can be described objectively to study temporal and regional variations for a wide range of conditions associated with DSD. We would recommend that all cases of gonadectomy should be recorded in a registry such as the I-DSD Registry for the purpose of long-term audit and surveillance.
In the process of performing this study, we believe that we have identified simple, yet effective, clinical benchmarks, that is, the number of individuals undergoing gonadectomy prior to review at a specialist centre or the age of presentation compared to the age at gonadectomy. Benchmarking represents a continual and collaborative effort at measuring and comparing the results of key processes that are directly relevant to patients with an eventual aim of improving clinical outcomes (26). Gaps in the professional development of service providers have been reported before (27,28) and it is possible that targeting simple observations such as these leads to greater local awareness of the pros and cons of early gonadectomy. It is likely that there is an optimal balance to be achieved between timely gonadectomy and a well-informed patient or parent and further exploration of these measures may allow the identification of this optimal range. Identification of benchmarks that target specific points in the care pathway of the patient with gonadectomy may lead to improvement of this pathway but this will require further exploration. For instance, most specialist centres that participate in the I-DSD Registry act as regional centres of referral and it is possible that greater outreach links and education programmes between such specialist centres and the referral catchment area may lead to greater informed decision making. Knowledge and awareness can be promoted locally through participating in joint clinics or local educational activities within the referral network and the availability of such opportunities has been previously reported by DSD centres to be variable (27).
There are, however, some limitations to using the number of individuals undergoing gonadectomy prior to review at a specialist centre as a clinical benchmark. The definition of a specialist centres was based on the centre participating in the I-DSD Registry and it is of course possible that the centre prior to presentation to the I-DSD centre was also an expert centre. Future work should examine the complete pathway of care of the patient who undergoes gonadectomy more comprehensively. With only 1 in 20 individuals undergoing gonadectomy prior to presentation to a specialist centre, the need for further research and prospective monitoring of timing of gonadectomy is important to confirm whether the findings in the current study were simply a chance finding.
It is possible that this study may suffer from some selection bias, as not every patient at a centre may have been included in the Registry. On the other hand, the structured manner of real-world data collection within the Registry, the size of the cohort and its potential to represent global practice were clear strengths. In addition, as with any large database, errors may occur with data entry; however, in all data that were analysed were checked with the centres at the time of the study to ensure it matched with the source data. Although, the investigators did not have recourse to source data, the data that have been collected have previously been reported to have a high degree of reliability (29). Comparison of the case mix that exists in the I-DSD Registry (30) to the range of cases that may exist in a specialist service (31) also shows remarkable similarities. For a greater understanding of the practice of gonadectomy, it would be helpful to have further information on gonadal position, postoperative complications, patient and parental decisional regret, complications of gonads that are left in situ, geographical and temporal differences in social attitudes, local health care resources and, very importantly, the extent of information provision to patients and parents. These aspects were beyond the scope of the Registry but deserve further study. The current study was aimed at studying individuals older than 16 years at the time of data collection. Data from younger individuals would also be of interest, particularly in conditions for which the risk of germ cell tumour is high early in life. A prospective study, condition-specific study would also allow more valuable information on condition-specific trends which has not been possible in the current study due to the small numbers.
To conclude, this global collaborative effort has clearly demonstrated that although the timing of gonadectomy depends on the condition and the age of the patient, there is considerable variation between centres. Objective parameters that have been identified as part of this work may serve as benchmarks of practice that can be prospectively monitored.