Celiac disease and its serological pattern in Saudi Arabia: a systematic review

Aim: To analyze the status of Celiac Disease (CD) and its Serological Pattern in Saudi Arabia (SA). Methods: Using a retrospective systematic literature review, the status of CD in SA was analyzed from various aspects. The related articles (74) were retrieved by database and journal search. Data were analyzed using the statistical package for social science (SPSS Inc). Results: The retrieved studies mostly represent multi heterogeneity in terms of journals (42 journal), centers (23 center), mode (12 type), fields (21 discipline). Most publications were found in: Saudi Medical Journal (12 articles), King Saud University, Riyadh (21), cross sectional (18), retrospective (18), Case report (18), Diabetes mellitus(T1DM) (8); while down syndrome (DS), short status (SS), skeletal health and CD prevalence had 3, 5, 6 and 7 articles respectively. Female to male ratio was 1.95/1. Histological evaluation was introduced early1980s followed by serology in 1990s. Biopsy showed the highest usage (in 57 studies), endomysial-antibody (EMA) (34), tissue-transglutaminase-antibodies (tTG-Ab) (29), anti-gluten-antibodies (AGA) (18), anti-reticulinantibodies (ARA) (10). Publications started in 1990 with stable annual number (1-3) until it peaked (10 articles), in 2012, 2016 and 2017. Conclusion: The current study represents the first and only Systematic Review concerning the status of CD in SA and its serological pattern. We found one review evaluating skeletal health in CD in SA and one Meta-analysis concerning polymorphism association with CD, but no meta-analysis were found concerning the relationship between CD and other conditions such as T1DM, SS, DS and CD prevalence in SA, which will be our plan for separate publications.


Introduction
Celiac disease (CD) was first described in the second century AD by Aretaeus of Cappadocia [1], who used the Greek word "koeliakos", which means "suffering of the bowels". Only in 1888 AD did Samuel Gee [2] give it the classical clinical description. Its cause was explained by the Dutch Pediatrician "Willem K. Dick" who recognized an association between the consumption of bread and cereals and the relapsing diarrhea [3] and that the health of celiac children dramatically improved when wheat, rye and barley were unavailable during the 2nd World War but relapsed at the end of the war when the consumption of wheat flour started afresh in the Netherlands [4].
At the Interlaken meeting of the European Society of Paediatric Gastroenterology and Nutrition (ESPGAN) society in 1969 [5], three diagnostic criteria for celiac disease were proposed, and were further enunciated at the second symposium on celiac disease in 1974 [6], namely: * Structurally abnormal jejunal mucosa by taking a gluten-containing diet.
* Clear improvement of villous structure by a gluten-free diet.
* Deterioration of the mucosa during challenge.
In 1971, Seah and coworkers observed binding of antibodies in CD patient sera with rodent liver and kidney sections and identified antireticulin antibodies (ARA) [7]. CD serology started initially in the 1970s and 1980s using anti-gluten antibodies (AGA) immunofluorescece followed by ELISA respectively [8,9]. Chorzelski and colleagues (in 1983) described the production of endomysial antibody (EMA) in patients with CD and dermatitis herpetiformis [10]. AGAs also exist in healthy individuals (up to 10%) and in nonceliac-enteropathies, thus, diagnostic guidelines have recommended that AGA tests should be abandoned for routine diagnosis [11].
In 1990, the ESPGAN revised criteria for diagnosis of celiac disease [12] considered the finding of the circulating antibodies (AGA-IgA, ARA, and EMA) during diagnosis and their disappearance during a gluten free diet (GFD) as valuable to the diagnosis, and emphasized on tow diagnostic criteria with no need for further confirmation: firstly "on the appearance of flat small intestinal mucosa with the histological features of hyperplastic villous atrophy, while the patient is eating adequate amounts of gluten; and secondly on unequivocal and full clinical remission after withdrawal of gluten from the diet".
In 1989, A simple and flexible approach for diagnosis of celiac disease was suggested by Guandalini S, et al. [13].
In 1992, Marsh introduced a grading scheme to classify the morphologic spectrum of the mucosa of untreated CD patients [14]. Then, Marsh-Oberhuber classification [15] modified some of Marsh's parameters. In 1997, tissue transglutaminase tTG2 was identified as CD-specific autoantigen [16], which allowed the development of ELISA-based tTGA tests.
The prevalence of CD, which has been underestimated, is now considered one of the most common genetic disorders in the West with a prevalence of 1%-2.67% [17][18][19]. The literature regarding CD has expanded significantly in the Middle Eastern (ME) and North African (NA) countries during the last 30 years which are mostly epidemiological [20]. Many studies exist on celiac disease (CD) in Saudi Arabia , but data from these investigations have not yet been compiled. Here, studies on CD in SA were systematically analyzed, classified, characterized and their serological pattern was discussed to highlight the current status of CD in SA, with the aim that the results of this study will contribute in promotion of national guidelines for organized SD research and management.

Strategy for systematic search and study selection
As shown in the PRISMA flow-diagram ( Figure 1); using "celiac disease in Saudi Arabia", "celiac disease in Saudi children" and "prevalence of celiac disease in Saudi Arabia" as key words, an electronic literature search was conducted via PubMed (US National Library of Medicine, with no specific period), Ovid, EBSCO and scholar Google. The retrieved studies (articles) were designed as twelve groups: three groups (according to the three key words) by PubMed search and three groups by EBSCO, three groups by Ovid and three groups by scholar Google. Additional group of articles were obtained through the library of king Fahd research centre of King Abdulaziz University. In addition, we obtained (with thanks) the publications (two articles and one meeting abstract) related to "celiac disease in Saudi Arabia" directly from the editorial department of two local journals (Saudi Journal of Internal Medicine and Journal of King Abdulaziz University Medical Science). All the retrieved articles were checked for matching (duplication) between groups and intra groups via their titles, author(s) and year of publication. The obtained pooled studies (after matching between groups and intra groups) were checked by reading their titles and abstracts, and the full texts for the possibly relevant publications. Articles that were concerned with celiac disease in Saudi Arabia were included. All forms of studies were included (original studies, reviews, guidelines, surveys, questionnaire, letter to the editor, comments and correction; pediatric, adult, both pediatric and adult). Global reviews with author(s) from Saudi centers were also included. The selected (included) articles were arranged in four groups: references of both adult and pediatric subjects, references of adult subjects, references of pediatric subjects and references of animal models. Data from each article of this classification (as four groups) was recorded using statistical package for social science (SPSS Inc), Version 20. Chicago. The recorded data from each article comprised: author(s), date of publication, journal, date of the study, duration, place of the study, single or multi centres, number of patients, female/ male ratio, type of serology and whether with or without biopsy, type of study (cross sectional, retrospective, prospective, case report, letter to the editor, genetic, HLA, Case-control, meta-analysis, review, pilot prospective, animal experiment, correction to previous publication), pattern of study (general pattern, prevalence, genetic, association (with diabetes mellitus , with autoimmune disease(s), with short stature and association with small bowel (SB) damage), animal model, clinical and/or lab characterization (presentation), corrigendum to previous publication, cardiac assessment , diet (GFD) assessment, skeletal assessment, assessment of knowledge, general assessment, effect of breast feeding on autistic CD and effect of thyroidism.

Strategy for studies' grouping
Age wise, according to Al-Agha, et al. 2015 [33], puberty is defined with a cut-off level of 10 years in females, and 12 years in males; while the term "children and adolescent" denotes "aged one to 18 years" [33,39,40]. Thus, age wise, articles were divided into 3 groups: pediatric (<12 years in male, <10 years in female or male with females), adults with adolescents (>12 years) and pediatric with adults (and/or adolescents) (> 1 year (or 1-18 years)).
The reference section comprises the included articles  but not the excluded articles. The included references were serially arranged in two groups: the PubMed references (with PubMed numbers) followed by the non-PubMed references (without PubMed numbers). The PubMed references were further arranged in two subgroups: the "pediatric with adults (and/or adolescents)" group followed by references of "adult with adolescents", as no recorded pediatric references were found. The non-PubMed references were further arranged in three subgroups: "pediatric with adults (and/or adolescents)" group followed by references of "adult with adolescents" then the "pediatric" subgroup. References in each subgroup were arranged in descending chronological order.
According to PubMed classification, children are individuals aged 0-18 years, while adults are those above 18 years, other definitions of age-groups existed in which individuals aged 19-21 years have been included as children [95,96] indicating that the age group definitions were not always clear. Here, studies which clearly stated that they are concerned with children (without adolescents) or adults were considered accordingly, otherwise they were considered of both children and adults (and/or adolescents).

Statistical analysis
Analysis of data was performed using statistical package for social science (SPSS Inc), Version 20. Chicago. The results were illustrated in tabulated form and diagrams showing comparisons and frequencies of variables.  By EBSCO search using: • "Celiac disease in Saudi Arabia" (N= 33).
• "prevalence of Celiac disease in Saudi Arabia" (N= 12) All the EBSCO articles were also PubMed, that match with the previous PubMed articles.
By search in scholar Google (advanced) using: • "celiac disease in Saudi children" (N=40) • "celiac disease in Saudi Arabia" (N=9) • "prevalence of celiac disease in Saudi Arabia"/ N=Nil Only eight studies were new non-PubMed articles Fifteen articles were obtained through the library of king Fahd research centre of King Abdulaziz University. Two articles and one meeting abstract, related to "celiac disease in Saudi Arabia", were obtained directly from the editorial department of two local journals (Saudi Journal of Internal Medicine and Journal of King Abdulaziz University Medical Science).
All the retrieved references (articles) were checked for matching intra and between groups via their titles, author(s) and year of publication. Following the matching process,153 pooled articles were obtained. 82 articles were excluded that were not concerned with celiac disease in Saudi Arabia; they were 36 not celiac articles and 46 articles not related to Saudi Arabia. Three additional non-PubMed studies were obtained from the references of the included articles: Two pediatric and one adult.
II. Female to male ratio: Gender information was found in 53 articles, while missing [not specified (NS) (Table 2). Articles with "female more than male" were five time (45/8) those with "male more than female" ( Table 2). The ratio of the total female (1731) over the total male (884) was 1.95 /1 (Table 3). IV. Cohort: In this study, cohort is meant to be the number of human subjects that were involved in each study whether patients of CD (for different evaluations) or of other conditions (for CD association), normal subjects (for CD prevalence), normal controls (in the case control studies) and participants in surveys.
Information about cohort were found in 66 articles, while missing [not specified (NS) or not applicable (NA)] in 8 articles (2 animal model, 2 comments, 1 correction, 2 reviews, 1 meta-analysis) (   (2), four centers with 3 studies each, and fourteen centers (with one study each) which included two centers from outside the kingdom: the Townsville Hospital [29], and Baghdad University [28] concerning comments on another article from SA (Figure 4).

VII. Types versus fields (disciplines) of studies:
Twelve types were found. Most publications were found as cross sectional, retrospective and case report (18 each). Followed by case-control 6, prospective 4, review 2, letter to the editor 2, animal experiment 2, genetic 1, Metaanalysis 1, Pilot prospective 1, corrigendum to previous publication 1.  • Fourteen centers with one study each • 0ne center with two studies • Four centers with three studies each • 0ne center with ten studies • 0ne center with fourteen studies • 0ne center with twenty-one studies • Descriptive Statistics The 74 studies involved 21 disciplines: case report (in 17 articles), association with diabetes mellitus (8), prevalence (7), CD and skeletal (6), genetic (5), association with short stature (5), clinical and/or lab characterization (presentation) (5), association with autoimmune disease(s) (3), prevalence in Down syndrome (3), pattern of CD (1 questionnaire) (3), Diet (2). One article for each of the following ten disciplines: general, Breast fed and CD and autistic, association with cardiac, corrigendum to previous publication, effect of thyroidism, CD and pregnancy, genetic and dental, psychiatric, Extra-abdominal conditions pattern, quality of life (questionnaire).
Distribution of types versus fields (disciplines) and articles are shown in Table 2 Table 7).
Mode of combination between histology and serology in each article: Different patterns of cooperation between biopsy and serology in each article are shown in Table 8. The concerned information was found in 63 articles, while missing in 11 articles (2 animal models, 4 review and questionnaire, 2 comments, one survey, one correction and one with no details). Biopsy was used in 57 studies according to the following pattern of serological cooperation: eleven articles [22,30,3 3,35,39,40,50,62,65,83,88] of tTG-IgA and biopsy; eleven articles [21, [72,93,94] of Biopsy and improvement by GFD; two article [44,85] of EMA, tTG Ab, AGA, ARA and Biopsy; two articles [69,73] of AGA and biopsy; one article [89] of AGA, EMA-IgA, and biopsy; two articles [51,74] of AGA,ARA and biopsy; one article [75] of AGA, EMA-IgA, and biopsy; one article [89] of biopsy and no improvement by GFD. Biopsy was not used in 7 studies according to the following pattern of serological cooperation: two articles [64,78] of tTG-Ab, one article [43] of EMA, one article [71] of EMA and ARA, one article [27] of EMA and tTG Ab, one article [51] of AGA and ARA, one article [60] of EMA, tTG Ab, AGA, ARA. The maximum usage (54%) was for tTG-Ab (30 with biopsy and 4 without) either alone or in combination with other serological markers; followed by 16% for CD proven by biopsy, 20% for markers without tTG-Ab [EMA,ARA,AGA with different combinations] (10 with biopsy and 3 without), 6% for Biopsy and GFD (3 improved with GFD and 1 no improvement); the minimum usage (4%) was for AGA-Ab without tTG-Ab (2 with biopsy and EMA or ARA, one with ARA, but never alone). (Table 8)

Chronological usage of serology and biopsy:
The pattern of the chronological usage of serology and biopsy in SA is shown in Table 9 and Figure 5. The two earliest studies of 1990 [93,94] indicated that CD proven by histology (with improvement by GFD) was the mode of diagnosis in 1980s. The study of Zawawi TH, et al. [91], that covered the period 1988-1994, indicated that all serology except tTG-Ab (EMA, AGA and ARA) were used during that period (1989)(1990)(1991)(1992)(1993)(1994). Also, there was no mention for tTG-Ab in the publications of 2001,2002 and 2003 [73,71 and 51 respectively]. The first mention of tTG-Ab was found in 2004 [86,87] indicating its definite presence in 2004.

Discussion
The current study is a statistical evaluation of all relevant literature on CD in SA, published up to March 2018 with no retrospective date limit. It represents the first and only Systematic Review concerning the  status of CD (and its serological pattern) in SA. However, we only found two studies from SA, analyzing the literature concerning specific fields of CD: one Meta-analysis of 11 global case-control studies concerning the genetic polymorphism association with CD [58], and one review (by literature electronic search using the MEDLINE and EMBASE databases) evaluating the global skeletal health (bone mineral density (BMD) and fracture risk) in CD from 1996 to 2010 [38], in addition to a general review about global CD, authored from SA [34].
In the 74 included articles, the ratio of the total female to total male was almost the double (1.95/1). Globally, CD is more common in female (double or triple) than in male, with a possible attribution to the higher frequency of the necessary HLADQ 2/ 8 alleles in female than male [97,98].
We found different rates of CD in different studies in SA. CD prevalence in SA is ranging from 1.5 % to 3% in four studies [23,30,63,64]. The global prevalence is 1% (with a range of 0.5-1.6%) [99][100][101][102], with a high prevalence (5.6%) among the" Saharawi refugees of Berber-Arabic origin" in North Africa [103]. One of the articles [26] studied the seroprevalence of celiac disease in at-risk subjects (T1DM, T1DM + Hypothyroidism, Hypothyroidism, Short stature, Failure to thrive, Abdominal pain, Chronic diarrhea, Anaemia, Others) which was 18.4%. Concerning the global at-risk groups (including Down and Turner syndromes, type 1 diabetes, and autoimmune thyroid disease) the prevalence is ranging from 5% to 10% [104].
Concerning the association of CD with other conditions, we found strongly variable rates of CD in different conditions in different studies. The prevalence of CD cases in patients with Rickets differed from 6.87% to 38.4% [42,36 respectively], while in patients with Low BMD differed from 43% to 70% [84,90 respectively]. Globally, low bone mineral density (BMD) at both the spine and hip in celiac patients was also reported [105][106][107]. It was also reported that celiac disease may cause rickets [108,109]; even, congenital rickets may be caused by untreated maternal CD [110]. In India, rickets rate within patients suffering of celiac and Type1 Diabetes Mellitus was reported to be 20.8% [111].
One article described a case of CD among 20 cases (5%) of APS type 1 [44], and one article described an association (42.8%) of CD and juvenile RA [52]. Cases of APS type 2 which were associated with celiac disease were reported [115,116]. In Italy, prevalence of CD in RA was nil [117]. In UK, the overall prevalence of celiac disease in RA is 0.63% [118], while an arthritis is present in 26% of CD patients, with difference between patients on a regular diet (41%) and patients on a gluten-free diet (21.6%) [119].
Two studies of questionnaire survey to the families of CD patients. One showed a significant negative socio-economic impact of Gluten-Free Diet (GFD) on children with CD & their families [25]. And one questionnaire about the adherence to GFD [32] showed that 60% of the involved children were reported to be strictly adherent to GFD; the younger age at diagnosis and the shorter duration since the diagnosis the better adherence rate.
Adherence to Gluten-Free Diet can be evaluated by different methods: serology [tTG-Ab (TG2) test], rebiopsy and dietitian interviews or/and questionnaire; however, antibodies take 1.5-3 months to normalize, but the finally, complete histological resolution may not occur until 2 years of GFD [120]. "Celiac Dietary Adherence Test" (CDAT), a 7-question survey that allows a standardized evaluation of adherence and is superior to the tTG-Ab test [121]. Patient interview for monitoring dietary compliance has been proven to be more sensitive than serology [122].
Globally, the degree of strict adherence to a GFD is strongly variable with a a range of 42% to 91% [123]. Less adherence is among adolescents, and adults diagnosed in childhood; due to lack of knowledge, poor availability and/or labeling of gluten free products, and difficulties in obtaining and/or identifying gluten-free food when dining out [124].
The information, within the articles, indicate for the usage of four serological tests (AGA, ARA, EMA and tTG-Abs), in addition to the biopsy test. Globally, these investigations were described in the fall of the second half of the last century [7][8][9][10][11][12][13][14][15][16] (see the introduction). However, with the start of this century (in 2004) a new test (DGPAs) was introduced for the detection of deamidated gliadin peptide Abs, which is promising according to its sensitivity and specificity [125] and appears to be possibly more useful than tTG in children below the age of 7 years and/or in IgA-immunocompromised patients [126,127]. We found no mention (within the articles) to this new serological test. Recently (in 2013), the first electrochemical immunosensor for DGPAs was described, as a good alternative to the traditional ELISA kits [128].
HLA typing is also missing in the retrieve articles. This test identifies whether a patient possesses the haplotypes DQ2 or DQ8, the absence of which excludes the possibility of CD [132]. HLA-DQ2 and/or HLA-DQ8 typing is useful in both and diagnosis and epidemiological  Table 9. Chronological usage pattern of serology and biopsy evaluation. Among the global population, HLA-DQ2 and/or HLA-DQ8 typing is present in 98.6% of patients with CD, and in 40% of the general population who do not have the diagnosis of CD [133].

Conclusion
Over the last decades, the availability of both specific and sensitive serological markers for CD, accompanied with technical advances have led to improvement in the frequency and accuracy of data on CD in SA. Since 1990, there have been multiple studies concerning the status of CD from many parts of SA, using many types of studies and covering a wide range of conditions, for both the general population and at-risk patients. Most results showed strongly variable rates of CD, among the general population and among at-risk patients, in different studies in SA. However, some conditions still need to be evaluated, such as the small intestinal bacterial overgrowth (SIBO) among patients with celiac disease unresponsive to gluten-free diet (GFD [134,135]. IgA-anti-tTG and AGA showed the highest and the lowest utilization rate respectively, due to their level of sensitivity and specificity which were high [sensitivity (98%) and specificity {98%} in tTTG-Ab, in comparison with AGA [(85%), {90%}]. Although EMA and ARA possess high sensitivity and specificity, they have some drawbacks including low reproducibility and high inter-observer variability, as both based on indirect immunofluorescence which requires an individual reading of each sample under a fluorescent microscope. The literature from SA missed the new currently used test for deamidated gliadin antibodies (DGP-IgG and DGP-IgA), and the DQ2 or DQ8 typing which is employed for both epidemiology and diagnosis, only when negative , to rule out the possibility of CD diagnosis. 3. Evaluation of DQ2/ DQ8 typing for both diagnosis and epidemiology.

Evaluation of Small intestinal bacterial overgrowth (SIBO) among
patients with celiac disease unresponsive to gluten-free diet (GFD).
These recommendations will be our plan for future separate publications.

Ethical approval
The collected data were part of a retrospective literature review; thus, a written ethical approval was not obtained before commencing the study.

Disclosures
Both authors have read and approved this manuscript. The current study was not funded or supported by any drug company. This paper is unique and is not under consideration by any other publication and has not been published elsewhere.