Research question

The main PICO question (acronym for Population-Intervention-Comparator-Outcome) of the initial systematic review was: What sort of endocrinologic follow-up is recommended to monitor the onset of endocrinologic diseases in NKX2-1-related disorders?. It was designed to serve as a guide for the literature search. Later, it was divided into three research questions to cover all the aspects related to the detection, diagnosis, treatment, and follow-up of endocrine alterations in NKX2-1-RD. The questions related to initial screening and diagnosis have been addressed in a prior study. Here, we focus on the following question: What are the best procedures for treatment and follow-up of endocrine diseases in patients with NKX2-1-related disorders?. The protocol of the systematic review was previously registered in the PROSPERO repository with the identification CRD42022341011.

Eligibility criteria

Specific inclusion criteria were employed to select the relevant studies. The population criteria included patients of all ages with genetic confirmation of NKX2-1-RD (pathogenic variants in TTF-1/NKX2-1 or deletion in the 14q13.3 chromosome) and hypothyroidism. Non-human studies and patients without BHC genetic confirmation were excluded. Given the rare condition, any comparator was used. The intervention of interest to treat endocrine alterations was defined as LT4 treatment. The follow-up interventions to monitor the evolution of the disease were the quantification of thyroid hormone levels in serum and hearing tests and ultrasound examinations as additional clinical evaluations. Any other type of intervention related to hypothyroidism but not to NKX2-1-RD was excluded. The outcomes based on treatment were the effectiveness of LT4 treatment and adverse effects. The outcomes related to follow-up were established as neurodevelopmental improvements, puberty development, bone, metabolic, and cardiovascular health, patient and professional education, quality of life, and self-management in adulthood. The study designs included were primary studies, systematic reviews, and randomized controlled trials (RCTs). Narrative reviews, conference articles, editorials, letters to the editor, and studies whose full text could not be obtained were excluded.

Search strategy

To review the scientific evidence, a literature search was carried out in reference databases such as PubMed, Embase, the Cochrane Library, MEDLINE (Ovid), PsyncINFO, CINAHL, the TRIP medical database, and the Health Technology Assessment (HTA) database. Subject-specific databases were included to retrieve data about rare disease resources such as Orphanet, EURORDIS, NORD, RARE-Best Practices, and Gene Reviews. We searched clinical trial registries, such as CENTRAL and the International Clinical Trials Registry Platform (ICTRP). The initial search covered the period from January 2002 to May 2022, with an update in July 2023. The records that have been collected since 2002 pertain to the initial discovery of the pathogenic variants of the NKX2-1 gene as the cause of BHC [12]. Both controlled language (descriptors) and free terminology were used to search for studies. The initial strategy was carried out in MEDLINE (Ovid) and later adapted to each database’s syntax (S2 File).

Study screening and data extraction

The references from the searches were imported into the software application Covidence (https://www.covidence.org/) for the screening process, where the duplicates were removed. Two authors independently filtered the references according to the inclusion and exclusion criteria (BCH and JDOE). The first selection was carried out by title and abstract, and the second one by full-text screening. Both reviewers reached an agreement in the case of discrepancies. The disagreements were resolved by a third researcher (JL). The data were extracted by two independent authors (BCH and JDOE) and recorded in Excel spreadsheets. Specific details of the studies (author, year of publication, study type, and location) and specific clinical data of each patient (age, sex, genetic tests, thyroid, neurological, and respiratory affections, age at onset of hypothyroidism, thyroid gland alteration, treatment regimen, effectiveness of treatment, adverse events, and follow-up) were extracted. Missing data points were labeled with ’NA’ (Not Available) to indicate their absence in the dataset. These data points were not included in the final analysis. For analyses requiring complete datasets, cases with missing data were excluded listwise (complete-case analysis) to avoid introducing bias due to imputation. Sensitivity analyses were conducted to assess the impact of excluding these cases on the study’s overall findings.

Quality assessment

The quality of the studies was assessed by two authors (BCH and JDOE) in an independent way. Disagreements between individual judgments were resolved by discussion to reach a consensus. The quality assessment was carried out using specific tools for observational studies, as the included references were classified as case reports, case series, and cohort studies. The tool developed by Murad et al. [13] was used for case reports and case series. This tool allows us to assess the methodological quality of the studies with a low body of evidence through eight items categorized into four domains (selection, ascertainment, causality, and reporting). According to the criteria of the tool, the quality of articles was rated by summing the scores of the eight binary responses (0 = high risk of bias, 1 = low risk of bias) into an aggregate score. The score ranges from 0 (very poor quality) to 8 (very high quality) [14]. The items of ascertainment, causality, and reporting were adapted to the specific clinical setting of the research question. The Newcastle-Ottawa Scale (NOS) was used to assess the cohort studies [15]. It uses three categories (selection, exposure, and comparability) with eight items to identify good-quality decisions using stars. One star can be given to each item in the selection (maximum of 4 starts) and outcome (maximum of 3 stars) categories, and a maximum of two for comparability. No studies were excluded on low-quality grounds. The Grade of Recommendation, Assessment, Development, and Evaluation System (GRADE) was used to evaluate the quality of the evidence provided by the studies. This system sets four categories (high, moderate, low, and very low), considering five factors: risk of bias, inconsistency, indirectness, imprecision, and publication bias [16].

Study selection

In the initial search, 458 records were identified. After eliminating duplicates, 453 potentially relevant studies remained. Title and abstract filters were applied, resulting in 104 studies for further assessment. Based on inclusion and exclusion criteria, 42 studies were included after full-text screening. The PRISMA flowchart in Fig 1 illustrates the screening process. The list of included and excluded articles with the reasons for exclusion, after full-text screening, is shown in S3 File. The excluded articles were classified according to the first criterion that did not fit the PICO format of the defined research question.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. PRISMA flowchart for treatment and follow-up of endocrine diseases in patients with NKX2-1-RD.

Illustration of the study selection process followed according to the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) guidelines. The flowchart depicts the number of records identified from the initial search, duplicates removed, and the number of studies excluded at each screening stage based on title and abstract filters, as well as full-text screening.

https://doi.org/10.1371/journal.pone.0309064.g001

Study characteristics

Table 1 shows the 42 studies analyzed. Among these, 50% were case series (n = 21), 47.6% were case reports (n = 20), and only 2.4% constituted cohorts (one study falling into this category). The rarity of the clinical condition under investigation resulted in the absence of randomized control trials or systematic reviews. Geographically, 61.9% were conducted in Europe (n = 26), 19% in America (n = 8), 16.7% in Asia (n = 7), and only one in Australia.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. List of included references for treatment and follow-up of endocrine diseases in patients with NKX2-1-RD.

This table provides details such as author, year, location, and study type of the final included articles.

https://doi.org/10.1371/journal.pone.0309064.t001

Patients demographics

The study included a total of 110 patients diagnosed with NKX2-1-RD and confirmed hypothyroidism. 53.6% were female, with a mean age of 15.2 years, ranging from 28 days to 64 years. The pathogenic variants and NKX2-1 deletions were frameshift 41 (37.3%), missense 38 (34.6%), NKX2-1 gene deletion 15 (13.6%), splicing 12 (10.9%), and nonsense variants 4 (3.6%).

In 84 patients (76.3%), sequencing, including next-generation sequencing (NGS) and whole exome sequencing (WES), was used to identify the NKX2-1 gene variants. Other methods employed include comparative genomic hybridization (CGH) array (11 patients, 10%), fluorescence in situ hybridization (FISH) (2 patients, 1.8%), and karyotyping analysis (1 patient, 0.9%). In 12 patients (11%), the specific genetic technique used for deletion or variant identification was not reported. De novo mutations were identified in 31.8% of patients (n = 35).

CH, or subclinical hypothyroidism, was the most common endocrine alteration, affecting 60% of the patients (n = 66). Two pregnant women showed gestational hypothyroidism [17, 18]. 36.4% (n = 40) had overt hypothyroidism. At least one patient had euthyroid-sick syndrome [19].

Some patients showed other endocrine alterations, along with hypothyroidism. Two patients had dysfunctional pituitary glands [20, 21], with one having a mildly hypoplastic pituitary gland and a Rathke’s cyst and the other presenting with an empty sella. In another study, a cystic pituitary mass was detected along with low prolactin levels and macroadenoma [22]. A female CH patient had undetectable gonadotrophins, neonatal transient hyperinsulinism, growth hormone and cortisol deficiencies. Neonatal transient hyperinsulinism manifested on the 18th day, characterized by a biochemical finding of insulin at 10 mU/L accompanied by hypoglycemia necessitating treatment with diazoxide and chlorothiazide [23]. A second patient had a cortisol deficiency [19]. A male patient with hypothyroidism, low testosterone, and luteinizing hormone had severe hypogonadism as an adult [21]. A second male patient with subclinical hypothyroidism had growth hormone and serum insulin-like growth factor-I (IGF-I) deficiency [23].

NKX2-1-RD affects the endocrine, neurological, and respiratory systems. In this study, 60.9% (n = 67) reported having the complete triad of the disorder. 28.2% of the patients (n = 31) only had endocrine and neurological alterations, while endocrine and respiratory involvement were reduced to 8.2% (n = 9). In three patients, only endocrine affectation was reported without more details (2.7%). It should be highlighted that chorea and hypotonia were the most dominant neurological disorders diagnosed in the first years of life. Neonatal respiratory distress was the most common postnatal respiratory alteration. NKX2-1-RD patients often have dysmorphic features, congenital anomalies, or developmental delay. The severity of the dysfunction determines whether it is detected in the first year or in childhood. Several patients in our study had learning impairments and behavioral disorders [24, 25]. Facial dysmorphism, palatal cleft, dental agenesis, fetal finger pads, a prominent forehead, and flaring nostrils were most visible in young patients [20, 26–28]. Some brain malformations included corpus callosum agenesis and dysgenesis, mild cerebellar atrophy, glioblastoma, hippocampus dysmorphism, and microcephaly [24, 29–33]. One patient had supratentorial leukoencephalopathy [29], and another patient had multiple birth defects [34]. 6 (5.4%) study participants died from severe clinical complications [18, 19, 35–38]. A detailed summary of the baseline characteristics of the included patients is provided in S4 File.

Treatment regimen for endocrine alterations in patients with NKX2-1-RD

In our systematic review, only 12 patients (10.9%) from 11 studies were reported to receive LT4 treatment after confirming CH [27, 28, 33, 39–42], subclinical hypothyroidism [24, 30], and overt hypothyroidism [43, 44]. Eight of these patients had altered neonatal screening, and four had overt hypothyroidism. In patients with overt hypothyroidism, the age of treatment initiation was 18 months, 21 months, 2 years, and 4 years. Screening centers had different hypothyroidism threshold values, causing uncertainty and disagreement. Many studies did not provide detailed information on neonatal screening or subsequent diagnostic tests, resulting in fluctuations in the initiation and dosage of LT4 treatment among patients and studies. Based on these data, the mean and median LT4 dose at the start of treatment were 52 and 37.5 μg/day, respectively. The mean age of treatment was 10.7 months, with a median of 35 days (range: 8 days to 4 years). Three patients received weight-based doses [30, 40, 44], with variations in the initial dose due to differences in age at onset and severity of hypothyroidism. The patients who had an altered neonatal screening started the LT4 treatment at neonatal age, in the first month of life, to normalize the hormone levels rapidly [28, 33, 39, 40]. Among the patients with a normal neonatal screening or those in whom it was not reported, they were diagnosed later, so the LT4 treatment started at older ages with the consequent physical and clinical alterations [30, 31, 41, 44]. Additional treatment regimen details can be found in S5 File.

Follow-up of endocrine alterations in patients with NKX2-1-RD

The follow-up period after hypothyroidism confirmation was reported in 11 of the 12 patients who received LT4 treatment. They constitute only 10% of the total number of patients in this study. TSH levels were used most commonly for monitoring thyroid function alterations. They were monitored from the beginning of treatment with LT4 for regular periods of time to control thyroid hormone fluctuations. The discontinuation of LT4 treatment and dose adjustments to regulate thyroid hormones were common strategies. Due to the complexity of NKX2-1-RD, the frequency of the follow-up also depended on the resolution of the hypothyroidism and the severity of other clinical manifestations such as neurological and/or respiratory alterations. In the 11 patients who received LT4 treatment and had a follow-up period, they showed positive responses. The dose of LT4 was gradually increased or diminished in order to normalize thyroid function, along with regular monitoring.

As stated in the preceding paragraph, we found important differences and a lack of data in the initial dosage of treatment. However, some patients reached a similar dose during the follow-up period, although the starting dose showed differences [30, 42, 44]. In two patients, the treatment was suspended at 4 years [27] and 18 months [30] of age because of a permanent stabilization of TSH levels. However, data on treatment effectiveness were not reported in a study [33]. During the thyroid hormone monitoring, a patient showed thyroid function normalization at day 30 of life, although the LT4 dose was gradually increased during childhood to maintain thyroid levels in the normal range. However, he continued with a bone development delay detected at 2 years. The remaining follicle-stimulating hormone, luteinizing hormone, and testosterone hormone values were adequate for the developmental age [28].

Similarly, a male patient was diagnosed with subclinical hypothyroidism along with growth hormone (GH) deficiency [30]. During the follow-up, thyroid function was normalized due to LT4 treatment, but neurological alterations and growth delays persisted. LT4 and GH treatments were administered to the patient, who showed clinical improvements. A neurological treatment was not considered due to the decrease in the severity of symptoms. Notably, one patient showed psychomotor delay and hypotonia, although euthyroidism was maintained after LT4 therapy [41]. This patient experienced discontinuation of treatment for four weeks, leading to disproportionate increases in TSH levels due to resistance. A second patient suffered frequent drop attacks if LT4 treatment was suspended, although he experienced stable euthyroidism. The drop attacks resolved upon treatment restoration [44]. However, it is important to acknowledge that these episodes may potentially be associated with the onset of chorea in an infant learning to walk. While the link between the disappearance of drop attacks and LT4 treatment is viewed as highly speculative, it is worth noting that the resolution of drop attacks could also be attributed to levodopa treatment. No studies reported adverse events related to the treatment. As anticipated, LT4 treatment improved stuttering, fatigue, and weight gain [39]. To track treatment efficacy and cover all follow-up procedures in NKX2-1-RD patients, hearing and ultrasound tests were included in the research question. Finally, this review found no evidence for these interventions. Detailed information on patient monitoring is provided in S5 File.

Complementary treatments for neurological and respiratory alterations in patients with NKX2-1-RD

A total of 40% of the patients received treatment for chorea and other movement disorders. Levodopa-carbidopa was the most commonly used treatment, showing effectiveness in 57% of the patients. However, side effects were reported in 24% of the patients, including dyskinesia, falls, sedation, dizziness, fatigue, drowsiness, drooling, freezing of gait, digestive intolerance, insomnia, nervousness, and asthma attacks [17, 34, 45–47]. Regarding respiratory alterations, 14% of the patients received treatment. The main treatments included surfactant for newborns with respiratory distress, antibiotics for infections, and oxygen. These treatments were effective in 12% of cases. Side effects were reported in 9% of patients and were primarily associated with prolonged steroid use, leading to conditions such as Cushing’s syndrome, adrenal insufficiency, osteoporosis, hypokalemia, and reduced bone density. Additionally, one patient presented with neutropenia as a result of hydroxychloroquine administration [18].

Quality assessment of the included studies

In this systematic review, the quality of each study was evaluated on the basis of the scoring systems. According to the assessment tools, 73.2% (n = 30) of case series and case reports were considered to have poor methodological quality, 19.5% (n = 8) medium quality, and 7.3% (n = 3) good quality. The cohort study was assessed as of poor quality [48]. The scores for each item and the total score for the studies are listed in S6 File.

Quality assessment of the evidence

Due to the limited number of evaluated patients and the variability of the data, a quantitative synthesis of the results was not feasible in this systematic review. The inadequate study design downgrades the initial quality assessment. The estimation of the risk of bias, inconsistency, indirectness, imprecision, and publication bias was made qualitatively. Firstly, the data showed high variability due to the different treatment regimens among patients in dosage, age at initiation, follow-up time, and dose adjustment. Moreover, the lack of reported data on the severity of thyroid alterations and the hormone levels at neonatal screening and/or diagnosis tests makes it difficult to establish LT4 treatment ranges in correlation with the previous patients’ parameters. Secondly, the indirectness of the study was considered low as the available evidence was directed at the initial research question. The imprecision of the results was high due to the small sample size and the lack of reporting of results of interest. The literature searches were designed to be sensitive enough to retrieve studies related to the research question. It is possible that some of them have been lost, but the results reported are similar in terms of not giving information about the values of the treatment dose, the follow-up, the age at onset, and the effectiveness of the treatment. Despite this, they show that they consistently align with the current available evidence on treatment strategies for NKX2-1-RD. Regarding publication bias, it was not possible to perform an objective assessment due to the lack of information from unpublished studies. Therefore, the quality of the evidence collected in this systematic review can be considered low for the previous factors analyzed.