Introduction

Cerebrotendinous xanthomatosis (CTX; OMIM 213700) is a rare, autosomal-recessive, lipid storage disease, with more than 400 cases reported worldwide [1,2,3,4,5]. It is caused by pathogenic variants in the CYP27A1 gene, leading to sterol 27-hydroxylase activity deficiency. This results in reduced primary bile acid synthesis, particularly severely depleted chenodeoxycholic acid (CDCA) levels, abnormal deposition of cholesterol and cholestanol in the tissues, and increased excretion of bile alcohols in urine [2, 3, 6, 7].

CTX is a severe, chronic, progressive disorder requiring life-long care. Several hallmark signs are seen in both paediatric patients (e.g. neonatal cholestatic jaundice and early psychiatric symptoms) [8,9,10,11], and occuring throughout life (e.g. infantile-onset chronic diarrhoea, juvenile cataracts, tendon xanthomas, intellectual disability and progressive neurological deterioration) [3, 8]. However, the type, onset and severity of symptoms vary considerably between patients [5, 8].

Alongside the hallmark clinical features, biochemical and molecular genetic tests are typically used to diagnose CTX through detection of increased levels of plasma cholestanol and/or identification of CYP27A1 pathogenic variants [12]. However, a diagnostic delay of approximately 20–25 years has been reported [3, 8, 13]. This is thought to reflect the difficulty in recognising signs of CTX and lack of awareness surrounding the condition, often leading to misdiagnoses [3]. Increased understanding of best diagnostic practices is therefore needed to facilitate earlier recognition, diagnosis and treatment initiation.

First line treatment uses exogenous CDCA to restore the biochemical abnormalities in CTX [3], thereby improving clinical outcomes. Other licensed and unlicensed therapies that have been used alone or in combination include cholic acid [14], 3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase inhibitors (statins) [14,15,16,17,18,19,20], low-density lipoprotein (LDL) apheresis [21,22,23], and ursodeoxycholic acid (despite not being effective in CTX) [24,25,26,27,28]. Treatment is therefore inconsistent and more specific guidelines on best treatment practices are required. Furthermore, the most appropriate tests for monitoring treatment efficacy and timing for these, is yet to be confirmed [4, 29, 30].

If left untreated, patients experience poorer prognosis, progressive, irreversible neurological damage and reduced life expectancy [14, 31]. Whilst early diagnosis and CDCA treatment initiation can reverse or prevent disease progression/deterioration [10, 31,32,33,34], there is still differing opinion about the best time to start treatment, to ensure the best prognosis [3, 35].

There are currently no published guidelines focussing specifically on diagnosis, treatment and management of patients with CTX, despite evidence that early diagnosis and long-term treatment can improve symptoms and prognosis [31,32,33, 36]. The Delphi method is a systematic and robust methodology that uses iterative rounds of questionnaires to elicit expert consensus opinion [37]. In this modified Delphi panel, conducted between April 2019 and March 2020, we sought to establish consensus on questions regarding best practices for the diagnosis, treatment and management of patients with CTX.

Methods

Study design

This study used a modified Delphi method; whilst classical Delphi studies continue until consensus is achieved for all questions [38], this study included three rounds to avoid questionnaire attrition and to acknowledge that some questions may not reach consensus, even after several rounds (Fig. 1).

Fig. 1
figure 1

Delphi study design. *Total number of questions asked. For some Likert scale questions, multiple options were given, with each analysed separately to assess consensus (e.g. “During the early stages of treatment, paediatric patients should be monitored for the types of symptoms listed below 1–2 times per year. Symptoms: Central and peripheral nervous system; Ocular system; Enterohepatic system; Cognitive performance [e.g. learning difficulties]; Cardiovascular system; Skeletal system; Pulmonary system”, where consensus was assessed separately for each symptom). †For each round, questionnaires were left open for responses in SurveyMonkey until analyses of the results began

Full size image

To increase the number of experts who could answer questionnaires, an external agency (KH, BSi, AG and DS) independently coordinated the study and developed questionnaires, given the small number of CTX experts worldwide. To ensure relevance, questions were validated by an unpaid expert clinician (BSt) chosen by Leadiant Biosciences, who did not answer questionnaires. Responses remained anonymous to other panellists. Pooled results and individual comments were shared with panellists and Leadiant Biosciences.

Targeted literature review

A targeted literature review (TLR) was conducted to identify published literature to guide development of the Round 1 questionnaire (Fig. 1). Search terms used in the electronic databases are presented in Additional file 3: Table S1 and full eligibility criteria are shown in Additional file 4: Table S2.

Delphi panellists

In addition to those who scoped the study design (MTD, AF and AV), experts invited to be panellists were identified based on the first and last authors of relevant full-texts in the TLR, or authors of case studies determined to be relevant at the abstract sift stage of the TLR (Additional file 1: Study methodology). Only the Delphi coordinators and AF, who distributed the email invitations, were aware of panellists’ identities. Invitation emails were sent to 18 potential panellists, with 12 agreeing to participate, 10 completing Rounds 1 and 2, and 9 in Round 3.

Questionnaire development and distribution

Relevant full-text articles identified from the TLR (Additional file 5: Table S3) were prioritised to inform five sections within the Round 1 questionnaire: diagnosis, treatment, monitoring, multidisciplinary care and prognosis (detailed methodology in Additional file 1: Study methodology and Additional file 2: SurveyMonkey® questionnaires). Questions that reached consensus in Rounds 1 or 2 were removed from the subsequent round. Questions that did not reach consensus were restated or rephrased in the subsequent round to increase the likelihood of achieving consensus (Fig. 1). The Delphi coordinators decided whether to rephrase or restate a question based on the responses in the previous round, comments from the panellists, and the advice of the expert validating the questionnaires (BSt). Round 2 and 3 questionnaires (Additional file 2: SurveyMonkey® questionnaires) were sent alongside individualised Microsoft PowerPoint® (Microsoft, Redmond, Washington) presentations summarising the pooled results of the previous round to encourage elicitation of consensus.

Question types and pre-specified consensus thresholds

Round 1 included six ‘profiling’ questions to understand panellists’ backgrounds, perspectives and CTX experience (Table 1 and Additional file 2: SurveyMonkey® questionnaires). The five main question types used and pre-specified consensus thresholds, are detailed in Table 1. Each question included “do not wish to answer (DNW)” or “insufficient expertise (IE)” options.

Table 1 Question types and consensus thresholds used in questionnaires
Full size table

Processing and synthesis of results

Results were exported from SurveyMonkey® and analysed in Microsoft Excel® 2016 (Microsoft, Redmond, Washington). Consensus was assessed by collating and calculating the response distribution. IE responses were excluded in the statistical analyses and DNW was considered neutral (detailed methodology in Additional file 1: Study methodology).

Results

Delphi study participation

Of the 12 experts that were sent the Round 1 questionnaire, 10 completed it (Fig. 1). Most panellists were neurologists, with other healthcare professions including geneticists and metabolic specialists, some with paediatric specialities. Most panellists were affiliated to a CTX specialist centre/department, with others from either a local or university hospital, based in a range of countries worldwide (Table 2). The majority had experience of treating > 5 patients with CTX within the past 10 years, with 40% (4/10) having cared for/treated adults only. The number of years for which panellists had been treating patients with CTX, varied (Table 2).

Table 2 Panellist demographics
Full size table

Questionnaire results

In Round 1, of the 22 questions that were used to assess consensus, 13 (59.1%) reached consensus. In Round 2, 32 questions were asked following Round 1 analyses, and 7 (21.9%) of these reached consensus. Following Round 2 analyses, 26 questions were asked in Round 3, with 1 (3.8%) reaching consensus. The results from all three rounds can be found in Table 3 (Likert scale questions), Table 4 (ranking questions) and Table 5 (proportion questions). The response distributions for the Likert scale, ranking and proportion questions can be found in Additional file 6: Tables S4–S6.

Table 3 Responses to Likert scale questions
Full size table
Table 4 Responses to ranking questions
Full size table
Table 5 Responses to proportion questions
Full size table

Diagnosis

Panellists agreed that symptoms presented by paediatric patients prior to a CTX diagnosis include chronic diarrhoea, bilateral juvenile cataracts and intellectual disability (e.g. learning difficulties). Panellists agreed that symptoms presented by adults prior to diagnosis are infantile-onset diarrhoea, childhood-onset cataracts, tendon xanthomas, psychiatric symptoms and neurological symptoms (peripheral neuropathy, cerebellar and pyramidal signs). Panellists also agreed that early-onset dementia presents in 25–49% of adults before diagnosis. It was agreed that movement disorders can be late CTX manifestations, however, CTX should be considered in the differential diagnosis of movement disorders, particularly in early-onset cases and when associated with other neurological and/or systemic features.

In terms of the most important indicator when considering a CTX diagnosis, CYP27A1 pathogenic variants were considered to be of greatest diagnostic value. When ranking the importance of tests/examinations used to confirm a CTX diagnosis, panellists agreed that genetic testing is most important, followed by determination of serum cholestanol levels. All panellists responded that patients always have elevated serum cholestanol at diagnosis and that measuring serum cholestanol is the diagnostic marker of choice. It was agreed that dried bloodspot (DBS) testing is the optimal method for screening of CTX in newborns.

Treatment

Panellists agreed that the most beneficial time to start CTX treatment is from birth, following a positive newborn screening test. Initiating treatment upon CTX diagnosis (with or without symptom onset) was agreed to be the next most beneficial option, followed by starting treatment upon symptom onset in diagnosed patients.

All panellists agreed that CDCA is a lifetime replacement therapy which may be capable of reversing the pathophysiological process in CTX, especially if initiated early in the disease process. The following ranking order was agreed upon when panellists were asked to consider therapy options that are effective for treating the underlying biochemical abnormalities in CTX: CDCA alone was ranked first, CDCA and HMG-CoA reductase inhibitor together was ranked second, and LDL apheresis was ranked last. No consensus was reached for other treatment options such as cholic acid alone.

Consensus was not reached on how useful available parameters for measuring treatment efficacy in patients with CTX are. In Round 1, serum cholestanol levels and clinical presentation/neurological examination were ranked as most useful by 30% and 50% of panellists respectively. In Round 2, serum cholestanol levels alone was most commonly ranked as the most useful parameter (50% of panellists agreed), whilst in Round 3, clinical presentation/neurological examination was most commonly ranked first (56% of panellists agreed). However, brain MRI was not ranked the most useful parameter for measuring treatment efficacy in any round.

The following order was agreed upon when panellists were asked to consider factors that have the greatest impact on treatment outcomes: age at diagnosis and treatment initiation was ranked first, the extent of neurological deterioration was ranked second and cholestanol levels at diagnosis was ranked last.

Monitoring

Panellists agreed that providing patients with support and intensive education improves treatment adherence. Panellists were in consensus disagreement that transcranial magnetic stimulation (TMS) is a useful tool for evaluating pyramidal function improvements in patients receiving CDCA. When asked about monitoring symptoms (during early stages of treatment as this is when dose adjustment may be necessary), panellists agreed that cognitive development and symptoms of the ocular, enterohepatic and central/peripheral nervous systems should be monitored in paediatric patients once or twice annually, and in adults annually. Additionally, symptoms of the cardiovascular/skeletal systems should be monitored in adults annually.

The following order was agreed upon when panellists were asked which examinations/tests for monitoring paediatric and adult patients receiving treatment are most useful: testing for cholestanol plasma concentration was ranked first and neurologic examination (and if necessary neuropsychologic examination) was ranked second. No consensus was reached on the ranking order for brain MRI, liver function tests and urinary bile alcohol concentration.

Panellists agreed that paediatric patients should undergo tests for plasma cholestanol concentration and liver functions once or twice annually, and neurologic (and if necessary neuropsychologic evaluation) examination twice annually. It was agreed that adults should undergo each of these tests once annually. Regarding the use of brain MRI for monitoring patients, there was no consensus on whether patients should have an MRI once annually. Comparing brain MRI and clinical evaluation, consensus was not reached on whether MRIs allow tracking of CTX disease progression with greater sensitivity than clinical scales and whether they should be used during follow-up.

Multidisciplinary care

Panellists agreed that neurologists and paediatricians/metabolic specialists should be involved in the diagnosis of, prescribing treatment to and follow-up of all patients with CTX. Agreement regarding the roles of other clinicians is presented in Table 3.

Prognosis

All panellists agreed that patients with CTX who start treatment after significant neurological pathology is established, have a worse prognosis compared to those who start treatment as early as possible. All experts responded that CDCA alone improves/stabilises prognosis in the majority of patients with CTX. When considering CDCA in combination with HMG-CoA reductase inhibitor without being presented with an option of CDCA alone, panellists agreed that this therapy option also improves/stabilises prognosis. However, panellists responded that CDCA alone is the preferred first line treatment compared to CDCA in combination with HMG-CoA reductase inhibitor for treating underlying biochemical abnormalities in CTX. Panellists disagreed that LDL apheresis improves/stabilises prognosis. No consensus was reached on whether therapy with cholic acid alone or in combination with HMG-CoA reductase inhibitor improves/stabilises prognosis. It was agreed that reducing plasma cholestanol concentrations slows down CTX progression.

Consensus was reached on some questions regarding the use of neurological imaging to indicate disease prognosis. Panellists agreed that patients showing MRI evidence of cerebellar vacuolation should be monitored more strictly over time as it is considered a poor prognostic marker. There was agreement that increased atrophy and/or signal alteration identified through brain MRIs may be present in those with deteriorating neurological symptoms. Additionally, panellists agreed that the absence of dentate nuclei signal alteration may indicate better prognosis. Consensus was not reached on whether brain MRI can be used to determine neurological stability.

Discussion

There are currently no standard guidelines on the diagnosis, treatment and management of patients with CTX. This Delphi study achieved consensus on aspects of care for paediatric and adult patients, from a group of experts.

In paediatric patients, chronic diarrhoea, bilateral juvenile cataracts and intellectual disability (e.g. learning difficulties) were identified as key symptoms, supported by available literature [3, 9]. For patients who are not diagnosed until adulthood, additional symptoms include tendon xanthomas, psychiatric and neurological symptoms. Presence of these signs should prompt clinicians to refer patients for further testing, and greater awareness of these typical symptoms may aid early diagnosis and treatment [3]. Whilst some signs asked about in these questions did not reach consensus, in some instances their presentation should still prompt further investigation. As clinicians may only see a small subset of patients, whose symptoms will likely vary from patients seen by other physicians due to the heterogenous nature of CTX, this may indicate why consensus was not achieved for these questions. For example, neonatal cholestatic jaundice can often be self-limiting [39, 40], and so the proportion of patients presenting to each expert with this symptom could vary, dependant on the age of patients at presentation. Nevertheless, when prolonged and without a specific diagnosis, neonatal jaundice should raise suspicion towards a CTX diagnosis given it can potentially cause irreversible liver damage if left untreated [8, 41].

Presence of biallelic CYP27A1 pathogenic variants was considered to be the indicator of greatest diagnostic value. This suggests that, whilst the presence of key symptoms may prompt further investigation [3], molecular analysis of the CYP27A1 gene should be considered the primary means for diagnosis, with whole exome sequencing showing great promise for increasing accurate diagnoses [42]. Results indicate that patients always have elevated serum cholestanol levels and that this should be considered an appropriate secondary means for investigating a diagnosis. However, the fact that in some atypical cases patients exhibit normal cholestanol levels [43], should be acknowledged. Furthermore, raised serum cholestanol levels have been described on occasion in patients with primary biliary cirrhosis and Niemann Pick type C, and very rarely in progressive familial intrahepatic cholestasis type 3 [44]. The fact that elevated cholestanol levels are sometimes observed in other conditions should therefore be recognised when considering this indicator in the investigation of a CTX diagnosis.

An alternative means for diagnosis, suggested as the solution for early identification of patients with CTX, is through newborn screening [45]. Several biomarkers have been described, with 5β-cholestane-3α,7α,12α,25-tetrol glucuronide (GlcA-tetrol) and the ratio of GlcA-tetrol to tauro-chenodeoxycholic acid (GlcA-tetrol/t-CDCA), highlighted as key candidate biomarkers for newborn screening [46, 47]. However, no national programme has yet been implemented [4]. It was agreed in this study that DBS testing is the optimal method for newborn screening and that starting treatment from birth, following a positive newborn screening test, is most beneficial. Supported by agreement that age at diagnosis and treatment initiation has the biggest impact on treatment outcomes, this indicates that suggestions for pilot screening studies should be taken forward [45]. However, this study did not explore the most appropriate biomarkers for CTX screening in newborns and so further research is needed to assess the best options for potential future newborn screening programmes [46].

Panellists agreed that providing support and intensive education to patients with CTX can improve treatment adherence, and that pre-marital genetic counselling should be recommended to high-risk populations. Such populations could, for example, include Moroccan Jews and the Druze community in the Middle East [48, 49]. Given the relatively high frequency of autosomal recessive diseases in these populations [50], the feasibility of providing counselling to these patients would need to be carefully considered.

There was agreement that CDCA is a lifetime replacement therapy that could reverse the pathophysiological process in CTX and improve/stabilise prognosis, and that it is the most effective therapeutic option available for treating underlying biochemical abnormalities, in line with recommendations in England that CDCA should be used in CTX treatment [51, 52]. Whilst panellists agreed that combination therapy with CDCA and HMG-CoA reductase inhibitor is also effective for treating CTX, CDCA alone was the preferred first line treatment. This aligns with literature where CDCA is considered to be standard of care for CTX [3, 33, 45]. Conversely, the lack of consensus about the effectiveness of cholic acid aligns with the paucity of evidence in the literature regarding its safety and efficacy, indicating there is not sufficient support for its routine use in CTX [4]. This is despite its recommendation in England as a second line treatment when CDCA is no longer tolerated or effective, and indication for bile acid synthesis disorders due to single enzyme defects including CTX in the USA [52, 53]. The consensus in this study that LDL apheresis is the least effective therapy option indicates that its use in CTX may not be appropriate.

Consensus was not reached on methods for measuring treatment efficacy. However, panellists agreed that cholestanol plasma concentration tests are most useful for monitoring treatment over time (for example, in facilitating dose changes), and should occur one to two times annually for paediatric patients and once annually for adults. Furthermore, panellists agreed that reducing plasma cholestanol concentrations slows down CTX progression. However, while serum cholestanol may be useful for evaluating treatment compliance [3], research on the impact of cholestanol levels, and whether cumulative cholestanol exposure correlates to disease progression, is lacking. Further research is needed to collect data that determine the usefulness of plasma cholestanol levels, or other metabolic precursors, for monitoring CTX.

This study emphasises the importance of initiating treatment early [32, 34]; if possible, prior to presentation of neurological symptoms, to ensure better prognosis. There was consensus surrounding neurological signs that should be monitored to determine prognosis; panellists agreed that evidence of cerebellar vacuolation, increased atrophy and/or signal alteration in brain MRIs can suggest worsening prognosis, whilst absence of dentate nuclei signal alteration in MRIs could indicate better prognosis. However, it is not clear what tools should be used for monitoring neurological signs, as demonstrated by the lack of consensus around some related questions in this study. Panellists did, however, agree that liver function tests and neurologic examination should be used to monitor all patients.

As a technique for eliciting expert consensus, the Delphi method has several advantages, allowing a variety of opinions to be gathered from a heterogenous sample of experts and for questionnaires to develop based on results and feedback in free-text responses. In our study, anonymity of panellists and their responses, and the use of an independent agency to coordinate the study, minimised external bias and maximised expert participation, as they did not have to act as coordinators. Validation of questions by one clinical expert who did not respond to the questionnaires ensured they were of high relevance and accuracy, maximising the usefulness of the output toward recommendations for the care of patients. Use of an online tool allowed responses to be quickly gathered from a group of geographically dispersed experts, and rapidly analysed.

However, there are some limitations to the Delphi method. Unlike standard questionnaires, Delphi studies require ongoing time commitment from panellists which can lead to questionnaire attrition. We minimised this by designing the study to only have 3 rounds, thereby not requiring panellists to commit to additional rounds if questions did not reach consensus, whilst still ensuring that all questions not achieving consensus were asked at least twice. There was therefore high panellist retention, with only one panellist dropping-out after Round 2. The Delphi study format also means that questions are interpreted by panellists without any explanation further to what is provided. However, having multiple survey rounds where panellists receive the pooled results from the previous round reduces the possible bias introduced through misinterpretation of questions, by providing an opportunity for re-interpretation in line with the group’s responses. Nevertheless, profiling questions, or questions that achieved consensus in Round 1, could have been subject to varied interpretation, presenting a possible limitation to this methodology. Furthermore, research indicates that social-psychological factors can cause experts with divergent views to feel pressure to conform [38]. Whilst AF, MD and AV received honoraria to attend the initial face-to-face study design meeting in September 2018, to limit bias introduced through sponsorship of this study by industry, panellists did not receive funding for participation in the Delphi panel. All panellists accepted authorship following completion of the study.

Given that CTX is a rare disease and there are consequently few experts worldwide, only a relatively small number of experts participated (n = 10 for Rounds 1 and 2; n = 9 for Round 3), representing a limitation of the study. Another limitiation of the study is the geographical spread of the panel, which was restricted as few experts worldwide from a relatively limited number of specialist centres were eligible to/agreed to participate. A larger and more diverse panel would ensure greater robustness and representativeness of the CTX population, although including experts from Europe, North America and Asia mitigated the limitation to some extent. A further limitation is that some experts had not been treating patients for very long and had treated relatively few patients in the past 10 years, which may be reflective of the small and dispersed population of patients with CTX, as a rare disease.

This study highlights several areas where expert opinion is aligned on the diagnosis, treatment and management of patients with CTX. However, there were some areas where consensus was not achieved; reasons that Delphi studies may not achieve higher consensus levels can include small sample sizes, divided clinical opinion, variability in disease presentation and importantly, a lack of data on particular topics being assessed. This study demonstrated that techniques for monitoring treatment efficacy and CTX progression require further investigation. In particular, given the importance of monitoring progression of neurological symptoms, further research to inform guidelines could be undertaken. Whilst the value of MRI biomarkers toward revealing disease prognosis was indicated, their use for measuring clinical improvement compared with clinical scales requires further investigation, e.g. benefits during presymptomatic stages or during slow disease evolution [29]. Further research also needs to identify what follow-up tests/examinations are appropriate based on the heterogenous clinical presentation of patients. Whilst offering a preliminary insight into preferred approaches for the care of patients with CTX, further data need to be collected to substantiate these findings, fill outstanding knowledge gaps, and inform best practices. Furthermore, this study collected opinions from healthcare professionals exclusively, and future research should therefore look to understand the impact of these decisions on patient/carer quality of life by gathering their opinions. Additionally, in light of agreement that provision of support and intensive education to patients can improve treatment adherence, opportunities to build on this in the future could be explored.

Conclusion

This Delphi study elicited consensus expert opinion on a number of factors relating to the diagnosis, treatment and management of patients with CTX. Results showed that, with a wide variety of symptoms throughout patients’ lifetimes, prompt diagnosis should be facilitated using genetic analyses or determination of serum cholestanol levels, or screening via DBS testing in newborns. Age at diagnosis and beginning treatment early (at birth, where possible) were considered to have the biggest impact on treatment outcomes and panellists agreed that early initiation of lifetime CDCA replacement therapy may considerably improve prognosis. No consensus was reached on the value of cholic acid therapy alone. Whilst results showed that patients should be monitored through plasma cholestanol concentration testing and neurologic examination once or twice annually, further research is needed regarding monitoring treatment/progression of the disease. This study highlights where further data are needed to inform best practices and provides an indication of preferred approaches for the care of patients with CTX.