Prevalence of Coagulopathy in Patients with Celiac Disease: A Single-Center Retrospective Case-Control Study

: Introduction: Despite being one of the most frequent chronic digestive diseases worldwide, with a prevalence of 1%, celiac disease (CD) remains severely underdiagnosed. Among the instruments used to improve its diagnostic rate, hematologic parameters have been proposed as screening tests to select patients with an increased probability of having CD. Assessment of coagulation is included in routine check-ups, and CD has been reported to be associated with coagulopathy. We aimed to assess if subtle changes in coagulation tests could be used in clinical practice to prompt testing for CD. Methods: We retrospectively recruited all patients with clinical suspicion for CD during a study period of 7 years (between 2015 and 2022), who were tested using IgA tissue transglutaminase (tTG) serology and serum total IgA (IgG tTG in case of IgA deﬁciency) and who underwent upper gastrointestinal endoscopy with multiple biopsy sampling of the duodenal bulb and distal duodenum. We stratiﬁed patients into three groups: newly diagnosed CD, gluten-free diet-treated CD, and non-CD controls. Results: Altogether, there were 133 CD patients (71 newly diagnosed, 62 GFD-treated) and 57 non-CD controls. Mean age and gender distribution were similar among the three groups: 43.3 years for newly diagnosed CD, 41.6 years for non-CD controls, and 44 years for GFD-treated CD patients, with a male gender distribution of 21.1%, 28%, and 24.1%, respectively. Among the included newly diagnosed CD patients, 14% had a prolonged INR. The mean INR was slightly higher in newly diagnosed CD patients, compared to GFD-treated CD patients and non-CD controls: 1.12 ± 0.30, 1.02 ± 0.83, and 1.00 ± 0.08, respectively ( p = 0.009). Consequently, prothrombin activity was slightly lower in newly diagnosed CD patients, compared to GFD-treated CD and non-CD controls: 94.9 ± 19.3%, 102.3 ± 12.8%, and 101.9 ± 15.15, respectively. Interestingly, after GFD, the mean INR and prothrombin activity of CD individuals reached a value similar to that of non-CD controls. Conclusions: Subtle changes in INR, deﬁned as a value within the normal range, but closer to the upper limit, could be an indicator of probability for CD.


Introduction
Celiac disease (CD), a systemic autoimmune disease that occurs upon gluten intake in genetically susceptible individuals, is one of the most underdiagnosed chronic digestive diseases worldwide.With a global prevalence of biopsy-proven CD of 0.7%, screening studies have shown an alarming rate of 75% CD underdiagnosis [1,2].Although instruments for CD diagnosis are commonly available and diagnostic algorithms are consistently reported in guidelines [3,4], the low detection rate is a consequence of several pitfalls in diagnosis [5]: missed screening of high-risk groups comprising associated autoimmune and non-autoimmune diseases (some of them long recognized-such as type 1 diabetes mellitus, autoimmune thyroid disease-and others more recently proposed, such as Sjögren's syndrome, pancreatic diseases) [6,7] and relatives of CD individuals [8][9][10]; missed detection of subtle mucosal changes in the duodenum in patients undergoing endoscopy for non-CD related reasons; insufficient knowledge about extraintestinal manifestations of CD among non-gastroenterologists [11,12]; missed testing for IgA deficiency [13]; challenges with serology, biopsy sampling, and histology; and, not least, misdiagnosing CD as irritable bowel syndrome [14].The consequences of not diagnosing CD can be substantial-from persistent nutritional deficiencies to malignant complications [15][16][17].
In order to retrieve more patients out of the submerged celiac iceberg, several casefinding strategies have been proposed over time.Among the instruments used to improve CD diagnostic rates, hematologic parameters have been evaluated as screening tests to select patients with an increased probability of having CD.Such routine hematology tests-complete blood count (CBC) and coagulation panel-are frequently performed on large-scale populations for various indications, including periodic medical check-ups, and changes in these parameters were analyzed with regard to their diagnostic performance and predictive value for newly diagnosed CD.Among them, there are several data on the use of red cell distribution width, on ratios of neutrophil to lymphocyte, platelet to lymphocyte, and others in CD screening, and on monitoring of dietary adherence in diagnosed CD individuals [18][19][20][21][22].In light of these results, we considered analyzing coagulation parameters in CD patients.
Assessment of coagulation is included in routine check-ups, and CD has been reported to be associated with coagulopathy.Both procoagulant and anticoagulant statuses have been described in CD [23]-Figure 1.The bleeding diathesis in CD is mainly theorized to be related to vitamin K deficiency and consequent alterations in vitamin K-dependent clotting factors, but also to the homology between factor XIIIA and tissue transglutaminase [24,25].On the other hand, a procoagulant state is depicted in CD, consisting in both arterial and venous thrombotic events, which can mark the clinical onset of CD [26,27].Although rare, thrombotic complications unmasking CD can be life-threatening, such as the case of a pulmonary embolism [28].The mechanisms behind hypercoagulability in CD are represented by hemorheological alterations, hyperhomocysteinemia, thrombocytosis (secondary to inflammation, iron deficiency, or functional hyposplenism), endothelial dysfunction, and the presence of thrombophilic antibodies [27,29,30].Regarding hemorrhagic manifestations of CD, they can represent the onset feature in some cases, and encompass a wide range of clinical presentations-from digestive hemorrhage (hematochezia, melena) to hemoptysis, hematoma, or hematuria-emphasizing Regarding hemorrhagic manifestations of CD, they can represent the onset feature in some cases, and encompass a wide range of clinical presentations-from digestive hemorrhage (hematochezia, melena) to hemoptysis, hematoma, or hematuria-emphasizing the need to increase awareness among other medical specialties to whom these patients might be referred [31][32][33][34].

Hemorrhage
As we routinely check for coagulation disorders in every patient who undergoes GI endoscopy in our unit, in anticipation of biopsy sampling (CBC with platelet count, prothrombin activity, international normalized ratio (INR), and sometimes activated partial thromboplastin time (aPTT)), we retrospectively assessed if subtle changes in coagulation tests were associated with a positive diagnosis of celiac disease.At the time of endoscopy, there was no hypothesis that these parameters might be indicative of CD, and we aimed to determine if altered coagulation tests could suggest the diagnosis of CD in a specific clinical setting and prompt specific testing.

Patient Recruitment and Study Procedures
We retrospectively recruited all adult patients with clinical suspicion of CD during a study period of 7 years (between 2015 and 2022), who were tested using IgA tissue transglutaminase (tTG) serology and serum total IgA (IgG tTG in case of IgA deficiency) and who underwent upper gastrointestinal endoscopy with multiple biopsy sampling of the duodenal bulb (one to two fragments) and distal duodenum (four fragments), in line with currently available guidelines [3,4].Tissue transglutaminase serology was tested using Celikey assay (Thermo Fisher Scientific, Uppsala, Sweden), with a cut-off for positivity of 10U.Endoscopy was performed by two experienced examiners using high-definition scopes (Olympus/Tokyo, Japan), and biopsy samples were formalin-fixed, paraffin-embedded, and stained with hematoxylin-eosin, with histopathologic changes being reported according to the Marsh-Oberhuber classification [35].Based on serology and histology, patients were stratified as either CD or non-CD controls.We also included gluten-free diet (GFD)-treated CD patients, who achieved mucosal healing on control endoscopy and had negative tTG serology, as a second control group.
CD patients were further stratified according to the type of presentation/diagnosis into three groups: (1) Gastrointestinal manifestations, representing the "classic" findings of CD: chronic diarrhea, bloating, weight loss, or signs and symptoms of malabsorption.
(2) Non-gastrointestinal manifestations: dermatitis herpetiformis, elevated liver enzymes of unknown etiology, infertility, iron deficiency anemia, osteoporosis in premenopausal women and young men, or oral ulcers.(3) A screen-detected strategy was applied in patients with following conditions: family history of CD, selective IgA deficiency, autoimmune thyroiditis, type 1 diabetes mellitus, autoimmune hepatitis, or Sjögren syndrome.
Routine laboratory work-ups including platelet count and coagulation tests were performed on all patients, using Beckman Coulter and ACL TOP 550 analyzers.We excluded patients with family or medical history of disorders of hemostasis, other potential causes for altered coagulation tests (anticoagulant use, chronic liver disease including liver cirrhosis, obstructive jaundice, hematologic disease, malabsorptive state-associated inflammatory bowel disease, common variable immune deficiency), and outliers, consisting of two patients with incoagulable INR.A prolonged INR was defined as values over 1.25.The normal range for prothrombin activity was considered 70-130% and 150-400 × 10 3 /mm 3 for platelet count.
Patient recruitment, the work-up performed, and assignment into one of the three groups are summarized in Figure 2. Using this methodology, we aimed to assess the coagulation profile of newly diagnosed CD individuals compared to non-CD controls and GFD-treated CD patients and to analyze if subtle changes in coagulation tests could be used in clinical practice to trigger testing for CD.

Statistics
Statistical analysis was carried out using SPSS 17 (Chicago, IL, USA) and Microsoft Excel Version 16.35 (Microsoft Corporation, USA).Quantitative variables are expressed as means with standard deviation or median with interquartile range, as appropriate.Comparison between groups was conducted using the Kruskal-Wallis test.An alpha value of 0.05 was chosen as the cut-off for statistical significance.

Ethics Approval
The study was approved by the Local Ethics Committee.All patients signed the research consent form included in their medical records, thereby providing written consent to participate in this study.

Statistics
Statistical analysis was carried out using SPSS 17 (Chicago, IL, USA) and Microsoft Excel Version 16.35 (Microsoft Corporation, USA).Quantitative variables are expressed as means with standard deviation or median with interquartile range, as appropriate.Comparison between groups was conducted using the Kruskal-Wallis test.An alpha value of 0.05 was chosen as the cut-off for statistical significance.

Ethics Approval
The study was approved by the Local Ethics Committee.All patients signed the research consent form included in their medical records, thereby providing written consent to participate in this study.

Results
Altogether, there were 71 newly diagnosed CD patients, 57 non-CD (controls), and 62 GFD-treated CD patients.Mean age and gender distribution were similar among the three groups: 43.3 years for newly diagnosed CD, 41.6 years for non-CD controls, and 44 years for GFD-treated CD patients, with a male gender distribution of 21.1%, 28%, and 24.1%, respectively.Among patients who obtained a CD diagnosis, 61.9% had gastrointestinal features suggestive of CD, 23.9% had non-gastrointestinal manifestations, and 14% were screen-detected (Table 1).A total of 7% of newly diagnosed CD patients had selective IgA deficiency.
Concerning the immunologic activity of the disease, 60.5% of patients in the CD-NEW group had tTG values over 10 times the upper limit of normal.All CD individuals had atrophic mucosal injury on duodenal biopsy samples.Mean time from diagnosis and initiation of gluten-free diet was 40.5 months for the CD-GFD group.
Prolonged INR was seen in 14% of newly diagnosed CD.The mean INR was slightly higher in newly diagnosed CD patients, compared to GFD-treated CD patients and non-CD controls: 1.12 ± 0.30, 1.02 ± 0.83, and 1.00 ± 0.08, respectively (p = 0.009) (Figure 3).Consequently, prothrombin activity was slightly lower in newly diagnosed CD patients, compared to GFDtreated CD and non-CD controls: 94.9 ± 19.3%, 102.3 ± 12.8%, and 101.9 ± 15.15, respectively, without reaching statistical significance (p = 0.100).Interestingly, after GFD, the mean INR and prothrombin activity of CD individuals reached a value similar to that of non-CD controls (Figure 3).

Discussion
CD is widely recognized as a clinical chameleon, which accounts for a significant delay in diagnosis and marked underdiagnosis of this disease.Although having a worldwide prevalence of 1%, the use of mass screening for CD is not recommended according to current guidelines.Therefore, clinicians need to adopt a case-finding strategy in order to increase the detection of CD in clinical practice.Despite good availability of diagnostic tools, there is still a need for detecting patients at risk for CD who warrant screening and confirmatory testing.
In order to increase the diagnostic rate of silent and atypical CD, several strategies have been tackled over the years: raising awareness about the disease and at-risk groups; using point-of-care tests; careful inspection of the duodenum to detect markers of villous Platelet counts of CD-NEW were significantly higher than those of controls or GFD-treated CD: 326.56 ± 121.03, 293.38 ± 75.72, and 266.32 ± 67.69, respectively (p = 0.022) (Figure 4).Thrombocytosis was seen in 18.3% of newly diagnosed CD patients.

Discussion
CD is widely recognized as a clinical chameleon, which accounts for a significant delay in diagnosis and marked underdiagnosis of this disease.Although having a worldwide prevalence of 1%, the use of mass screening for CD is not recommended according to current guidelines.Therefore, clinicians need to adopt a case-finding strategy in order

Discussion
CD is widely recognized as a clinical chameleon, which accounts for a significant delay in diagnosis and marked underdiagnosis of this disease.Although having a worldwide prevalence of 1%, the use of mass screening for CD is not recommended according to current guidelines.Therefore, clinicians need to adopt a case-finding strategy in order to increase the detection of CD in clinical practice.Despite good availability of diagnostic tools, there is still a need for detecting patients at risk for CD who warrant screening and confirmatory testing.
In order to increase the diagnostic rate of silent and atypical CD, several strategies have been tackled over the years: raising awareness about the disease and at-risk groups; using point-of-care tests; careful inspection of the duodenum to detect markers of villous atrophy, including bulb biopsy sampling to detect ultrashort CD; increasing adherence to the guideline-recommended biopsy protocol; and improving serologic, endoscopic, and histologic assessment [36][37][38][39][40][41].
Among the screening instruments proposed, routine blood tests that are performed very frequently in common practice have been evaluated as potential tools for selecting patients with the likelihood of having CD who would benefit from further testing.Studies reporting on ratios between complete blood count parameters (platelet/lymphocyte, neutrophil/lymphocyte, lymphocyte/monocyte) and red blood cell indices (red cell distribution width) have shown good diagnostic accuracy for CD (Table 2).A combination of these indices, defined as the systemic immune inflammation index, has also been evaluated for its diagnostic value in CD [42].On the contrary, others have recommended against using of these hematologic parameters as a screening for CD in specific patient groups, such as the study by Johnston et al. on insulin-dependent diabetes mellitus [43] or the one by Özdemir et al. in children with iron-deficiency anemia [44].Not least, there are some data on using these hematologic ratios as surrogate markers for monitoring CD patients [45][46][47].Changes in hematologic ratios and indices used for CD are related to anisocytosis and chronic inflammation [48].In the current study, we aimed to perform a similar assessment, but using coagulation parameters, which are also routinely available in clinical practice.There are already solid data published on hemorrhagic events associated with CD, so there might theoretically be a possibility of detecting early coagulopathy in some CD individuals who are not yet clinically apparent.We analyzed three patient groups-newly diagnosed CD, non-CD controls, and GFD-treated CD individuals-with respect to INR, prothrombin activity, and platelet count.Interestingly, 14% of patients in the CD-NEW group had prolonged INR, compared to 0% in non-CD controls and 3.2% in GFD-treated CD.Mean INR was slightly higher, but within the normal range, in CD-NEW compared to the other two groups, which had similar INR levels, meaning GFD-treated CD is, from a synthetic assessment of coagulation, almost the same as non-CD controls.Paralleling INR values, prothrombin activity was slightly lower, but still within the normal range, in CD-NEW compared to the two control groups.
Mean platelet count among sub-groups was remarkable in terms of significantly elevated values among CD-NEW, compared to controls: 326.56 ± 121.03 and 266.32 ± 67.69, respectively (p = 0.007).Thrombocytosis can occur secondary to anemia or hyposplenism in CD individuals, and resolution can be seen after institution of GFD.Considering their involvement in hemostasis, except for the obvious cases of thrombocytopenia or thrombocytosis, there seems to be no significant difference in platelet count among newly diagnosed celiac disease patients, as seen in our study subjects.
Despite the significant amount of data on hematology indices for CD diagnosis and follow-up, there are only a few studies on coagulation tests in CD (Table 3).In the study by Cavallaro et al. [54], which assessed the prevalence of impaired coagulation in 390 adults with untreated CD, prolonged prothrombin time (defined as INR ≥ 1.4) was seen in 18.5% of cases, higher than in our study lot.Altered coagulation was related to other markers of malabsorption, and 5.6% of patients required parenteral vitamin K substitution therapy, which reversed the coagulopathy in all cases.However, due to the low prevalence of coagulation disorders in their cross-sectional study, the authors did not support CD screening in patients with isolated coagulopathy as determined by routine testing.In pediatric CD, abnormal prothrombin time was seen in higher proportions: 27% of cases in the study by Sharma et al. [55], and 25.6% in the study by Ertekin et al. [56], with the proportion of deranged INR increasing with the severity of histologic lesions on biopsy samples [55].Other pediatric studies have found a lower prevalence of elevated INR, of 9.38% as reported by Dhankhar et al., similar to our results [57].Of note, despite the lower frequency of prolonged INR reported by Dhankhar et al. compared to the other studies, they found a high proportion of abnormal activated partial thromboplastin time among their subjects, of 31.25%,however without significant correlation with titer of anti-tissue transglutaminase antibodies or severity of mucosal damage on histology.Interestingly, in the report by Ertekin et al. [56], coagulopathy was only seen in typical forms of CD, while INR was normal in all patients with atypical CD, including screen-detected cases, while Cavallaro et al. [54] found a prolonged prothrombin time only in a small proportion (0.9%) of subclinical CD patients.The low prevalence of altered coagulation in atypical CD among published papers did not encourage authors to recommend screening in this setting, as the diagnostic yield would be too low.Thus, the detection of subclinical impairment in blood clotting through routine coagulation tests might be a good opportunity to identify clinically inapparent CD and perform additional diagnostic tests to confirm or rule out CD.Moreover, coagulation impairment in a CD patient is with high probability the result of vitamin K malabsorption, and, extrapolating, the severity of the coagulation disorder should be directly related to the extent of small bowel mucosal damage.Therefore, in a patient with CD and prolonged INR, an extensive assessment of nutritional deficiencies secondary to malabsorption should be performed.Besides vitamin K deficiency, patients with CD may also have low serum calcium (Ca) levels as a consequence of primary malabsorption or secondary to vitamin D malabsorption, and hypocalcemia is known to be a risk factor for coagulopathy.Various case reports described patients presenting with hypocalcemia and its complications, including coagulopathy, as the initial manifestation of celiac disease [58][59][60].Interestingly, some patients with symptomatic hypocalcemia had normal or elevated serum vitamin D levels, a finding which should therefore not argue against determination of serum Ca levels if clinical suspicion of hypocalcemia is high; however, before Ca supplementation, pseudohypocalcemia in patients with protein malabsorption should be excluded after correction for hypoalbuminemia.Thus, CD testing in the setting of mild coagulopathy should be triggered particularly when other clinical or biochemical signs or malabsorption are seen in suspected individuals.
We did not include patients with a personal or family history of hemostatic disorders that might be related to CD [61] in our study since in such a suggestive clinical scenario, further coagulation diagnostics would be warranted.We also excluded patients with co-existing diseases or treatments that might have affected the coagulation panel: advanced chronic liver disease, other digestive diseases associated with malabsorption, and anticoagulant treatments for hematologic diseases.
Despite the slight changes in INR in newly diagnosed CD individuals, reflecting a tendency toward increased bleeding risk, there are no data showing an increased rate of hemorrhagic complications from endoscopic biopsies required for diagnosis, except for isolated case reports [55,62].Moreover, there are already validated criteria for non-bioptic diagnosis in pediatric CD [63], and several studies have evaluated the feasibility of a biopsy-avoiding diagnostic strategy in adults as well [64,65], which would eliminate all potential complications associated with endoscopy and duodenal mucosal sampling.In a pediatric study with 111 biopsy-proven cases of CD, none of the children had bleeding following endoscopy; however, those with moderate and severe prolonged INR (2.51-5, and >5, respectively) were given preprocedural parenteral vitamin K.
The clinical significance of this mild coagulopathy associated with untreated CD seems limited.While studies reporting on impaired coagulation of CD patients did not mention overt bleeding among patients observed, despite some of them having markedly prolonged INR (>5 as in the report by Sharma et al. [55]), there are case reports of clinically significant hemorrhagic events [31].Moreover, severe coagulopathy has been reported in celiac crisis, which can occur during the course of disease in GFD non-adherent patients or as the initial presentation of the disease, and also in refractory CD [66,67].
While current guidelines lack any approach to coagulopathy in CD, a clinical practice report mentions abnormally easy bruising as the potential sole manifestation of CD and considers celiac, along with inflammatory bowel disease and chronic liver disease, worth being excluded when faced with a patient with isolated prolonged INR [68].Of note, the most recent guideline from the American College of Gastroenterology [69] discusses a possible clinical scenario when biopsy might not be safe because of cardiovascular or bleeding risk, and allows for a diagnosis of likely CD in symptomatic adults in whom scoping cannot be performed and who fulfill the non-biopsy pediatric criteria: tTG IgA in high titer (over 10 times the upper limit of normal) and positive anti-endomysial antibodies in a separate blood sample.
In summary, there is a potential benefit in triggering CD testing upon detecting INR values closer to the upper limit of normal, otherwise unexplained.As coagulation tests are performed on a large scale in routine practice, the impact on improving CD detection might be substantial.This should be validated with CD screening in population-based studies, in order to quantify the benefit of testing in low-risk settings.However, an elevated INR as a result of vitamin K malabsorption is a nonspecific finding and can be detected in a multitude of gastrointestinal conditions leading to malabsorptive states that were excluded from our study.There is a need for a more specific approach when raising the suspicion of CD in a clinical setting, and we consider that future research should address the correlation of hematologic abnormalities like anatomical and functional hyposplenism (low spleen volume, Howell-Jolly bodies on peripheral blood smear), iron, folate and vitamin B12 deficiency, and hyperhomocysteinemia with unexplained coagulation abnormalities and idiopathic thrombocytosis in a multivariate analysis, in order to define a scenario that could more accurately predict CD in clinical practice.
The current research has some limitations.Patient inclusion per protocol was conducted according to clinical suspicion for CD, so this may represent a selected patient population with a high probability of having the disease.Another limitation was the lack of data regarding anti-endomysial antibodies and HLA typing for DQ2/DQ8; however, these markers are not routinely used for CD diagnosis in the adult population.Not least, we did not report on aPTT values, as this test was not available in all patients.Because of the low adherence to follow-up, which is well recognized in adult CD [70], another limitation is that the GFD-treated CD group is not entirely constituted by the same patients as the CD-NEW group.However, our study had several strengths: first of all, we included a control group consisting of patients with clinical suspicion of CD, who underwent scoping and tTG serological testing to exclude CD autoimmunity; additionally, we included patients with GFD-treated CD, who achieved negative tTG titer and mucosal healing, to serve as a second control group, in order to emphasize the role of disease remission in the (partial) resolution of coagulation abnormalities.

Conclusions
Coagulation abnormalities, as detected by routine testing, are relatively commonly encountered in CD individuals.Subtle changes in INR, defined as a value within the normal range, but closer to the upper limit, might be an indicator of probability for CD, particularly in individuals with other clinical or biochemical signs of malabsorption.Future populationbased studies are needed in order to validate if an elevated INR without an identifiable cause could trigger testing for CD in clinical practice and if it would significantly improve the diagnostic rate, given that CD diagnosis is currently based on a case-finding strategy.

3 Figure 1 .
Figure 1.Coagulation imbalance in celiac disease, reflecting both the procoagulant and anticoagulant statuses.

Figure 1 .
Figure 1.Coagulation imbalance in celiac disease, reflecting both the procoagulant and anticoagulant statuses.

10 3 /
mm 3 for platelet count.Patient recruitment, the work-up performed, and assignment into one of the three groups are summarized in Figure2.Using this methodology, we aimed to assess the coagulation profile of newly diagnosed CD individuals compared to non-CD controls and GFD-treated CD patients and to analyze if subtle changes in coagulation tests could be used in clinical practice to trigger testing for CD.

Figure 2 .
Figure 2. Flowchart of patient recruitment and study procedures.Patients with clinical suspicion of CD underwent routine blood tests (including coagulation tests), immunological tests (serum tTG IgA + total IgA, IgG tTG in case of IgA deficiency), and upper GI endoscopy with biopsy protocol for CD; unconfirmed cases were included in the control group (non-CD), while confirmed cases (CD at diagnosis) were included in the group CD-NEW, and GFD-treated patients were included in the CD-GFD group.Abbreviations: CD = celiac disease; tTG = tissue transglutaminase; IgA = immunoglobulin A; IgG = immunoglobulin G; GI = gastrointestinal; GFD = gluten-free diet.

Figure 2 .
Figure 2. Flowchart of patient recruitment and study procedures.Patients with clinical suspicion of CD underwent routine blood tests (including coagulation tests), immunological tests (serum tTG IgA + total IgA, IgG tTG in case of IgA deficiency), and upper GI endoscopy with biopsy protocol for CD; unconfirmed cases were included in the control group (non-CD), while confirmed cases (CD at diagnosis) were included in the group CD-NEW, and GFD-treated patients were included in the CD-GFD group.Abbreviations: CD = celiac disease; tTG = tissue transglutaminase; IgA = immunoglobulin A; IgG = immunoglobulin G; GI = gastrointestinal; GFD = gluten-free diet.

Figure 3 .
Figure 3. Mean INR values among the three patient groups.Abbreviations: CD-NEW = newly diagnosed celiac disease patients; CD-GFD = gluten-free diet-treated celiac disease patients; non-CD controls = patients in whom celiac disease was excluded.

Figure 4 .
Figure 4. Mean platelet counts among the three patient groups.Abbreviations: CD-NEW = newly diagnosed celiac disease patients; CD-GFD = gluten-free diet-treated celiac disease patients; non-CD controls = patients in whom celiac disease was excluded.

Figure 3 .
Figure 3. Mean INR values among the three patient groups.Abbreviations: CD-NEW = newly diagnosed celiac disease patients; CD-GFD = gluten-free diet-treated celiac disease patients; non-CD controls = patients in whom celiac disease was excluded.

Figure 3 .
Figure 3. Mean INR values among the three patient groups.Abbreviations: CD-NEW = newly diagnosed celiac disease patients; CD-GFD = gluten-free diet-treated celiac disease patients; non-CD controls = patients in whom celiac disease was excluded.

Figure 4 .
Figure 4. Mean platelet counts among the three patient groups.Abbreviations: CD-NEW = newly diagnosed celiac disease patients; CD-GFD = gluten-free diet-treated celiac disease patients; non-CD controls = patients in whom celiac disease was excluded.

Figure 4 .
Figure 4. Mean platelet counts among the three patient groups.Abbreviations: CD-NEW = newly diagnosed celiac disease patients; CD-GFD = gluten-free diet-treated celiac disease patients; non-CD controls = patients in whom celiac disease was excluded.

Table 1 .
Patient characteristics among the three study groups.

Table 2 .
Studies reporting on hematologic ratios for CD diagnosis.

Table 3 .
Studies reporting on coagulation abnormalities in CD.