1. Introduction
Glycogen is the first energy source between meals and is especially abundant in the liver and muscles. In the liver, it serves as a glucose reserve for maintaining normoglycemia during fasting. Several enzymes act in the synthesis and degradation of glycogen, and a deficiency in any of these enzymes results in Glycogen Storage Diseases (GSDs), previously called glycogenosis, which causes abnormalities in the storage or use of glycogen [
1,
2].
GSDs comprise a group of inborn errors of metabolism (IEM) due to dysfunctions in glycogen metabolism causing multisystemic diseases that can present at any age from the neonatal period to adulthood [
1,
2,
3,
4,
5,
6]. The characteristics of GSDs depend on the abnormal site of glycogen metabolism. They are classified into numerical types that follow the historical order of their description, a system still widely used, at least until type VII [
2]. Recent reviews suggest 20 types and subtypes of GSDs, divided into hepatic, muscular, mixed (hepatomuscular), and others limited to specific systems (two cardiac forms and one cerebral). The estimated GSD incidence is 1 case per 20,000–43,000 live births, and 80% of hepatic GSDs are caused by types I, III, and IX [
4,
7,
8]. A summary of those with liver involvement is presented in
Table 1.
The first step in the diagnosis of GSDs is a characterization of the patient’s phenotype and clinical features [
5]. Although GSDs share similar clinical features to some extent, there is a wide spectrum of the phenotypic disease continuum [
4,
5]. The resulting phenotypic classes for each type of GSD can be used to aid diagnosis and treatment. These classes include hepatomegaly, hypoglycemia, intermittent myalgia, rhabdomyolysis, and hemolytic anemia [
3]. Thus, the diagnosis requires carefully investigating the patient’s medical history [
9]. Some metabolites inform the differential diagnosis of GSDs and can be useful for disease surveillance, including the measures of glucose, ketones, blood lactate, serum lipids and triglycerides, liver transaminases, uric acid, creatine kinase, and neutrophils count [
5]. Hypoglycemia is the hallmark of hepatic GSDs, and hepatomegaly is a cardinal manifestation of GSDs with liver involvement, except for GSD 0 [
4].
Historically, hepatic GSDs used to be confirmed with histological analysis via liver biopsy [
2], which should lead to a definitive diagnosis in most cases but are critically dependent on correct tissue processing [
10]. In recent years, invasive and technique-dependent exams, not always available in all services, have been progressively replaced by molecular studies in DNA extracted from peripheral blood lymphocytes or oral mucosa swabs, given the availability of target gene sequencing or multigenic panels, with the advantages of being a less invasive option for confirmatory diagnosis and often shortening the diagnostic odyssey [
5,
6,
10]. Therefore, the definitive diagnosis is a combination of clinical presentation, biochemical abnormalities, imaging, histology, the determination of enzyme activity in liver tissue, and molecular genetic analysis [
5,
9]. The limitations of genetic testing include the possibility of unidentified variants in tested genes and the individual’s symptoms having an etiology unrelated to variants in the genes tested [
5,
6].
4. Discussion
This study aimed to analyze whether the diagnosis of a specific GSD type could be made by the referring physicians on their clinical praxis. Therefore, cases were based on actual investigations instead of the construction of didactic examples. Thus, although theoretically some of the types and subtypes of GSD could be distinguished with a complete clinical and biochemical investigation, actual praxis showed that data collection tends to be incomplete, including missing biochemical data in some cases, besides the absence of other valuable tests for the diagnosis of hepatic GSDs such as glucose tolerance test or fasting blood glucose-lactate relationship and CT liver imaging. This shows the heterogeneity of investigation routines among physicians or medical specialties and the lack of laboratory support, even in centers for rare diseases. Moreover, a long and invasive investigation, including full biochemical analysis and hepatic biopsy for histological and/or enzymatic studies, seems to be replaced when a single less invasive testing approach (for example, oral swabs) based on molecular analysis is available as a first option test.
Regarding the group of experts, the relation between the hit rate and the time since graduation in medicine, the time of experience in GSD, and expertise in the number of patients with GSD treated during their careers did not present a linear result, making it impossible to assess the impact of these characteristics on diagnostic accuracy.
Given that the average assertiveness of the clinical forms was 47%, when analyzing the relation between the fill percentage and the average hit rate, it was observed that cases with less than 50% of data filling (135 forms) had an average accuracy of 43%, while patients with 50% of data filling or more (45 forms) had an average accuracy of 58%, representing a difference of 15 percentage points, favoring the identification of cases with data completion more than 50%.
Of the 45 cases, 21 (47%) had a percentage of correct answers greater than or equal to 50%, and 12 (27%) were correctly identified by all experts. Of the 24 (53%) cases with less than 50% accuracy, 14 (31%) were incorrectly identified by experts.
An average data filling of 29% was observed in cases not correctly identified by any experts, compared to an intermediate data filling of 38% in cases correctly identified by all experts. However, it is noted that the data filling was higher in the group that had between 1% and 49% accuracy. In this analysis, it is argued that despite the nine-percentage points difference in data filling between the cases that obtained 100% correct answers and 100% errors, it is not possible to create a direct relation between the data completion and the accurate diagnosis.
Despite the assertiveness analysis of the clinical cases not indicating a consistent relation between the data filling and accuracy, the availability of clinical and biochemical data reported in the patients were the main points of attention raised by the experts during the feedback. The experts’ feedback reinforces the possibility that determining factors that were inadequately investigated may have impacted the accuracy of the assessment.
When evaluating the cases with 100% correct and 100% incorrect evaluation and the comparison with the types of GSDs, it was noted that the GSD Ia stood out as the type best identified by the experts, followed by the GSD Ib. At the same time, GSD IX has the most significant number of unidentified cases, either because of greater knowledge reported by experts about GSD Ia or because of the lack of an investigation capable of differentiating this from other types. Theoretically, since an underlying metabolic pathway characterizes GSD I, it has the highest diagnostic probability based only on general clinical and biochemical aspects. In addition, because general clinical and biochemical aspects are shared by types IX and VI, which share an abnormality in the phosphorylase pathway, it is impossible to distinguish between them based on this information alone.
According to the assertiveness analysis by type and subtype, GSD Ia had the highest overall success rate per case and in all groups. This was also the type of GSD most present in the hypotheses and had the best accuracy when comparing the number of diagnostic hypotheses proposed with the hit rate.
Among the factors that may have influenced the clinical diagnosis by the specialist are the underage investigation for later manifestations of the disease, the clinical description, and incomplete complementary analysis, in addition to the overvaluation of a particular clinical finding (“false positive”) or the discarding the diagnosis in the absence of it (“false negative”), as well as the lack of knowledge of the clinical type by the specialist.
In the present study, the mean age at diagnosis was nine months for types Ia and Ib, 24 months for type VI, 27 months for type IXa, and 108 months for type III. Investigation was below the age of onset of later symptoms, such as hyperuricemia (typically starting after puberty) or adenomas (after the 2nd or 3rd decade) in type Ia, neutropenia (up to 9 years of age), and chronic inflammatory disease (diagnosed at an average age of 8.7 years) in type Ib, or progression to cirrhosis in types III and IXa (also after puberty) [
2,
7,
11]. Therefore, the expert interpretation of clinical data may be compromised by a lack of information regarding the natural history of each of these conditions.
Symptoms and signs such as hepatomegaly, hypertriglyceridemia, elevated liver transaminases levels, and hypoglycemia were widely investigated and were altered in more than 90% of the studied cases. These factors are relevant for diagnosing GSD; however, they are not determining factors for defining the subtype, as exemplified in the graph in
Figure 1.
On the other hand, signs such as neutropenia and CPK elevation have been poorly investigated in the present sample. They could have been better used to identify the GSD subtype without specific genetic testing. Such information suggests that the biochemical investigation of these cases was incomplete, being unhelpful for the correct clinical diagnosis of two of the three most frequent forms of hepatic and hepatomuscular GSD, as demonstrated in the graph of
Figure 3.
The overvaluation of a single clinical finding as strongly indicative of the disease when present, or discarding such a diagnosis when absent, may explain part of the low scores given by experts, such as neutropenia and chronic inflammatory bowel disease in type Ib. Similarly, the presence of progressive muscle symptoms (positive) or the absence of CPK elevation (false negative) may have made it difficult to identify cases with type III correctly.
The analysis of the experts’ proposed diagnostic hypotheses corroborates the statement, since of the seven cases diagnosed as GSD Ib by NGS, six were investigated for neutropenia, four positive and two negatives. Positive patients for neutropenia had an average accuracy of 90%, while negative cases for neutropenia had an average accuracy of only 15%. Some experts did not correctly identify the GSD Ib case without information on neutropenia. Likewise, two reports of neutropenia occurred in patients with subtypes Ia and IXa, with a mean accuracy of 10% (20% and 0%, respectively). These two cases add up to eight evaluations by the expert group, with seven mistakenly classified as GSD Ib.
CPK elevation was investigated and confirmed in four patients, but only two confirmed a final diagnosis of GSD III. The proportion of correct answers for the two cases of GSD III with CPK elevation was 90%, while the other two cases (Ia and IXa) were not correctly identified by any of the experts, with GSD III being the most common hypothesis in three of the seven evaluations. Another three GSD III cases confirmed via NGS were present in the project; however, without CPK investigation, the experts correctly identified none.
These same parameters were reaffirmed during the feedback analysis. Of the five cases in which the reason for an incorrect diagnostic hypothesis was attributed to confounding factors, the explanations refer to one case of elevated CPK, a characteristic suggestive of GSD III, but with a diagnosis confirmed via NGS of GSD Ia; two cases of neutropenia, a finding most suggestive of GSD Ib, however with a confirmed diagnosis of GSD Ia via NGS; and two patients with a confirmed diagnosis of GSD Ib, but without presenting neutropenia.
Indeed, the lack of a determining factor and the atypical presentations made it difficult for specialists in GSD to make an accurate diagnosis; however, just as relevant was the lack of knowledge of less frequent forms, such as GSD VI and GSD IX.
The only case of GSD VI present in this study was evaluated by three experts, being correctly identified by only one of them. Among the experts’ feedback, the only argument about this case justified the incorrect identification due to the rarity of this type of GSD and the absence of a specific clinical marker.
GSD IX is a less common type and presents subtypes following autosomal recessive or X-linked recessive inheritance, making its recognition more difficult without a clear family history. In the present study, patients with type IX had an accuracy rate of only 13%, compared to the overall average accuracy of 48%. It was also observed that, based on the nine cases of GSD IX that received feedback from the experts, in addition to the lack of data and professional experience with this specific type of GSD, the low prevalence and the similarity with other types of GSD were highlighted.
IEM are usually severe conditions, and the precise identification of the molecular basis of these diseases is essential for adequate patient treatment and genetic counseling [
12]. In addition to reducing costs and optimizing time, in these situations, molecular confirmation is critical in clinical practice to (a) differentiate distinct subtypes in a group that encompasses several clinically overlapping conditions, (b) establish a correct diagnosis aiming at clinical follow-up and management measures [
13], and (c) help in family planning and the genetic counseling of other family members [
14,
15]. Still, from the point of view of clinical research, it can serve to (d) describe new variants that cause or modulate the disease [
8,
16], (e) add helpful information in the discussion of the genotype–phenotype relationship and the reclassification of variants of uncertain significance [
17], or even (f) identify new genes that cause similar clinical conditions [
18].
5. Conclusions
It was observed in the present study that known characteristics such as hepatomegaly and hypoglycemia, typical symptoms of GSDs, are insufficient to determine the type and subtype, which makes a detailed investigation necessary when there is diagnostic suspicion. Without a suggestive factor characteristic of a specific type of GSD, such as neutropenia, neuromuscular symptoms, or a family history suggestive of X-linked inheritance, the most proposed diagnostic hypothesis was GSD Ia. This finding possibly occurs because it is the most prevalent and recognized type of GSD in clinical practice, which could cause GSD Ia to be the most frequent alternative with the highest success rate. However, characteristics considered determinants in identifying the specific types or subtypes of GSD, as cited above, are not exclusive, thus becoming factors that may have induced the experts to make a diagnostic error.
The experts also reinforced the importance of completing a full analysis during the feedback. However, only a modest increase in the correct identification of the type and subtype of GSD was observed when more clinical and biochemical data were available.
Another relevant factor is the moment of the clinical and biochemical evaluation of the natural history of the disease. The natural history of some types of GSD can lead to late symptoms that were not present at the time of diagnosis, which could have helped in a more correct diagnosis.
In addition to the conclusions presented, the study team aims, based on their most recent professional experience, to observe that not only in the specialized literature publications but also in clinical practice, NGS techniques are becoming increasingly cheaper and more accessible for experts and generalists, which has made its use increasingly frequent to the detriment of traditional single-target gene sequencing techniques or even biochemical tests as a first-line technique for diagnostic investigation. This reinforces a tendency for the diagnosis of these genetic conditions to migrate from the traditional sequence of “clinical evaluation followed by biochemical investigation followed by molecular confirmation” to the sequence “clinical evaluation followed by molecular investigation followed by biochemical confirmation (eventually ending in the second stage)” or, perhaps, even more paradoxical, to the sequence “molecular investigation then biochemical confirmation and finally clinical correlation”.
Given the challenges observed when seeking an accurate diagnosis based only on clinical symptoms and general biochemistry, this research proposes a paradigm shift via reinforcing the tendency to migrate from biochemical and enzymatic to molecular diagnosis due to its practicality, time, cost, and specificity. However, the importance of basic clinical assessment to improve clinical and laboratory data is critical for situations where molecular testing is inconclusive, such as interpreting variants of uncertain significance or the presence of a single variant in heterozygosity for a recessive condition; for these situations, functional tests might still be necessary to establish genotype–phenotype correlations. When laboratory data are previously collected in routine care, it reduces time and energy instead of patient recruitment for prospective data collection, such as in reverse phenotyping studies.