Risk factors associated with osteoporosis in a cohort of prospectively diagnosed adult coeliac patients

Abstract

Background

Up to 75% of patients with untreated coeliac disease (CD) present with osteopenia or osteoporosis. Guidelines do not express with certainty whether each patient with newly diagnosed CD should undergo a dual-energy x-ray absorptiometry (DEXA) scan.

Aim

The aim of this article is to evaluate the prevalence of bone mineral density (BMD) alterations at diagnosis and risk factors associated with osteoporosis.

Methods

A total of 214 adult patients (median age 38 years; female = 71.5%) newly diagnosed with CD underwent DEXA. The patients were divided into three groups: patients with normal BMD, those with osteopenia and those with osteoporosis. Clinical, histological and serological features were assessed and compared among the three groups. Logistic regression including relevant independent variables was performed.

Results

DEXA indicated that 39.7%, 42.5% and 17.8% of the CD patients had normal BMD, osteopenia and osteoporosis, respectively. Logistic regression indicated that features significantly associated with osteoporosis were male gender (OR 4.7; 95%CI 1.1 to 20.8), age ≥45 years (OR 6.5; 95% CI 1.3 to 32.2), underweight (OR 7.4; 95% CI 1.3 to 42.5) and greater histological damage (Marsh 3C; OR 5.8; 95% CI 1.4 to 24.1).

Conclusions

BMD alterations were found in 60.3% of newly diagnosed adult coeliac patients. Osteoporosis was significantly associated with age ≥45 years, male gender, underweight and Marsh 3C, suggesting that at CD diagnosis, a DEXA scan might be beneficial, particularly in these subgroups of patients.

Key summary

• Dual-energy x-ray absorptiometry (DEXA) scans have been recommended only for coeliac disease (CD) patients who are peri- or post-menopausal females or males over 55 years and those with overt malabsorption or a history of fragility fractures.

• In this study, bone mineral density alterations were found in 60.3% of newly diagnosed adult CD patients.

• Osteoporosis was significantly associated with age ≥45 years, male gender, underweight and Marsh 3C, suggesting that at CD diagnosis, a DEXA scan might be beneficial.

Introduction

Osteoporosis is a systemic skeletal disorder characterised by low bone mass and micro-architectural deterioration with increased bone fragility and consequently increased fracture risk. Osteopenia is a condition in which bone mineral density (BMD) is lower than normal but the risk of bone fractures is not significantly increased.^1,2 In select predisposed patients, this condition can progress to osteoporosis. Dual-energy x-ray absorptiometry (DEXA) is a non-invasive and accurate test for evaluating bone mass, and it is the gold standard for BMD assessment.³ It has been demonstrated that up to 75% of patients with coeliac disease (CD) suffer from bone mass loss (osteopenia or osteoporosis).^4–10

Despite the strong association between low BMD and CD, guidelines do not express with certainty whether each patient with a new diagnosis of CD should undergo a DEXA scan.^11–13 At present, DEXA investigations are suggested only for patients with CD who are peri- or post-menopausal females or males over 55 years and those with overt malabsorption or a history of fragility fractures.^12,14,15 This is likely because studies on BMD in CD patients have led to discrepant results, possibly due to their heterogeneous designs.^12,14 Previous evidence is conflicting owing to the retrospective nature of most relevant studies, which assessed children and adults, measured different skeletal sites and also included patients who had already adopted a gluten-free diet (GFD).¹⁵ Moreover, because of the well-known gender effects of osteoporosis in the general population, studies of CD patients are often designed to include an exclusively female population of mostly post-menopausal individuals.^16–18 All these variables make it difficult to understand the real prevalence of low BMD and the possible influences of clinical features related to this condition in newly diagnosed coeliac patients.

The aim of this study is to evaluate, in a cohort of consecutive newly diagnosed coeliac adults, the prevalence of BMD alterations at diagnosis and to verify the associated risk factors.

Materials and methods

Patients and study design

From January 2006 to December 2016, 253 patients (72% female) were consecutively diagnosed with CD at a university hospital in Rome, Italy. All the patients were Caucasian adults (median age 38 years, range 18–72 years) and were invited to undergo before starting a GFD a DEXA scan of the lumbar spine and femoral neck to investigate the presence of osteoporosis or osteopenia. On the basis of the DEXA results, the patients were divided into three groups: patients with normal BMD, those with osteopenia, and those with osteoporosis.

At the time of diagnosis, all the patients completed a clinical questionnaire detailing the presenting signs and symptoms, including their duration before diagnosis, autoimmune comorbidities and other relevant clinical data. Presenting sign/symptoms were categorised as follows: gastrointestinal (GI) symptoms (including diarrhoea, abdominal pain and distension or irritable bowel syndrome-like symptoms) and extra-intestinal manifestations (anaemia, nutritional deficiency, and gynaecologic or dermatological alterations).^19,20 Weight and height were also obtained to determine the body mass index (BMI). According to the World Health Organisation (WHO), normal weight was defined as BMI > 18.5, while underweight was defined as a BMI below this limit, and obesity was defined as BMI ≥ 30 in both genders. To evaluate the possible factors involved in BMD alterations, all the potential risk conditions traditionally associated with osteoporosis in the general population were investigated: alcohol intake (≥3 alcohol units (AU) daily), smoking habit (more than 10 cigarettes/day for at least two years), previous bone fractures and chronic use of drugs that potentially affect bone metabolism (such as corticosteroids and selective serotonin reuptake inhibitors (SSRIs).^1,2 In women, menopause and abnormal menstrual cycles (amenorrhea lasting more than six months) were assessed.

Before starting a GFD, all the patients underwent serological assessment that included the following measures: immunoglobulin A (IgA) class antibodies anti-transglutaminase (Ab anti-tTG) and anti-endomysium (EMA), total IgA, red blood cell count, ferritin, triglycerides, cholesterol and parathormone (PTH).

To verify the risk factors associated with BMD alterations, all the variables described, including serological results, were analysed and compared among the three groups.

The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a prior approval by the institution’s human research committee.

The study was approved by the local Ethics Committee Institutional Review Board, Sapienza University (9 November 2005). Written informed consent was obtained from each patient.

Diagnostic investigations

Endoscopic procedure and histological classification

For the CD diagnosis, gastroscopy using a flexible video-gastroscope (Olympus) and at least four biopsies obtained from the second part of the duodenum were performed in all patients when they were on a gluten-containing diet. The biopsies were examined by an expert pathologist after haematoxylin and eosin staining and immunohistochemical staining for CD3 counts. The pathologist was unaware of the patients’ clinical data; CD was diagnosed when duodenal biopsies showed intestinal villous atrophy (mild, moderate or severe) with crypt hyperplasia and intraepithelial lymphocytosis, according to the Marsh/Oberhuber classification.^21,22

Serological assays

EMA and Ab-tTG IgA were assessed in all patients at the time of diagnosis. An indirect immunofluorescence assay was used to detect the IgA EMA in monkey oesophagus sections. IgA Ab-tTG were assayed using an enzyme-linked immunosorbent assay kit commercially available from Eurospital (Trieste, Italy). Diagnosis of selected IgA deficiency was made with decreased serum levels of IgA (<7 mg/dl) with normal serum IgM and IgG levels. In case of IgA class-selected deficiency, TTG, EMA and deamidated gliadin peptides (anti-DGPs) IgG class were assessed. The other blood assays, such as ferritin, red blood cell counts, total cholesterol, triglycerides, PTH and albumin, were performed using standard laboratory techniques. Anaemia was defined as haemoglobin <12 g/100 ml in women and <13 g/100 ml in men.

DEXA

BMD in the lumbar spine and left femoral neck were measured by DEXA. The BMD values are expressed as standard deviation (SD) scores that compared the individual BMD measurements to those of young adults (T score). Based on the WHO criteria,^1,2 osteopenia was defined as T scores between −1.0 and −2.5, while osteoporosis was defined as T scores of −2.5 SD and below in at least one of the two examined sites (lumbar spine and/or femoral neck). A normal BMD value was defined as T scores of −1.0 and above in both examined sites.

Statistics

The data are expressed as medians and interquartile ranges or values and percentages. Comparisons among the different groups were performed using Fisher exact test and the Mann-Whitney U test, as appropriate. All variables that were statistically significant in the univariate analysis and a relevant independent variable (histological Marsh 3C damage) were included in the logistic regression model to detect features associated with low BMD at diagnosis. Odds ratios (ORs) and 95% confidence intervals (CIs) were used to describe the associations and were obtained by logistic regression analysis. In this analysis the ORs were adjusted for all the variables.

Two-tailed p values <0.05 were considered statistically significant. Statistical analyses were run using dedicated software (MedCalc Software, Mariakerke, Belgium, version 12.7.8).

Results

A total of 214 (84.6%) patients underwent DEXA scans and were included in the study (female = 71.5%, median age 38 years, range 18–72 years). The remaining 39 patients (15.4%) did not undergo DEXA examination and were excluded from the study. Their clinical features did not differ from those of the included patients.

The DEXA scan showed that 85 (39.7%) and 129 (60.3%) patients had normal or low BMD, respectively. Among the patients with low BMD, 91 (42.5%) had osteopenia and 38 (17.8%) had osteoporosis.

In this cohort of patients, only two female patients (aged 49 and 58 years) had a previous osteoporosis diagnosis. They took both oral and intramuscular vitamin D3 supplements intermittently over the last five years. At the time of CD diagnosis, they repeated the DEXA scan and were included in the study; the results from the most recent scan were included in the analysis.

Table 1 shows the features of the three groups. Compared with the patients with normal BMD, the patients with osteoporosis were older (p < 0.001) and less frequently female (p < 0.001), had a lower BMI and were more frequently underweight (p < 0.001), had previous bone fractures (p = 0.004), chronically used alcohol (p = 0.034), and had increased PTH values (p < 0.001). Regarding clinical presentation, the patients with osteoporosis presented with GI symptoms (p = 0.031) more frequently than the patients with normal BMD.

Table 1.

Clinical, serological and histological features of adult coeliac patients (CD) classified on the basis of bone mineral density (BMD) results.

	Pts with normal BMD n = 85 (39.7%)	Pts with osteopenia n = 91 (42.5%)	Pts with osteoporosis n = 38 (17.8%)	p valuea	p valueb	p valuec
Clinical features and lifestyle items
Female gender %	82.3	69.2	52.6	0.05	<0.001	0.10
Age, median (range)	38 (19–64)	36 (18–72)	48.5 (19–72)	0.79	<0.001	<0.001
BMI, kg/m2, median (range)	22.8 (16.8–38.2)	21.05 (16.7–32.8)	19.8 (16–28.5)	0.001	<0.001	0.16
Underweight (BMI ≤ 18.5 kg/m2) %	6.2	20.5	30.5	0.01	<0.001	0.25
GI symptoms leading to CD diagnosis %	38.8	57.1	65.8	0.02	0.03	0.84
Anaemia %	52.4	52.9	45.9	1	0.56	0.56
Presence of symptoms ≥5 years  before CD diagnosis %	38	32.4	35.5	0.98	1	0.80
Familiarity for coeliac disease %	17.8	17.9	11.1	1	0.42	0.43
Post-menopausal female %	4.8	3.3	23.7	0.71	<0.001	<0.001
Autoimmune comorbidities %	31.3	31.8	30.5	1	1	1
Previous bone fractures %	14.3	20.5	41.9	0.38	0.004	0.03
Drugs %	2.4	4.6	2.6	0.68	1	1
Smoking %	28	29	34.2	1	0.52	0.67
Alcohol (≥3 units daily) %	1.2	4.6	10.5	0.37	0.03	0.25
Serological parameters
Positivity to anti-EMA and/or  TTG IgA autoantibodies %	84.7	89.1	86.8	0.50	1	0.72
Increased PTH values %	19.7	31.1	54.3	0.13	<0.001	0.03
Ferritin, ng/ml, median (range)	11 (1–186)	9 (2–188)	9.5 (2-89)	0.59	0.47	0.86
Cholesterol, mg/ml, median (range)	171 (100–257)	168 (98–280)	168 (93-246)	0.41	0.36	0.73
Triglycerides, mg/ml, median (range)	64.5 (28–160)	70 (30–216)	76 (36-171)	0.47	0.06	0.18
Albumin, g/dl, median (range)	4.1 (3.1–5.0)	4.2 (3.0–5.6)	4.05 (1.6-4.9)	0.51	0.25	0.42
Histology
Marsh 3C %	42.3	54.9	55.3	0.23	0.24	0.70

Pts: patients; BMI: body mass index; EMA: Anti-endomysium immunoglobulin A antibodies; tTG: anti-transglutaminase immunoglobulin A antibodies; GI: gastrointestinal; PTH: parathormone.

^aNormal BMD vs osteopenia. ^bNormal BMD vs osteoporosis. ^cOsteopenia vs osteoporosis.

Compared with the patients with normal BMD, the patients with osteopenia more frequently had GI symptoms (p = 0.016), were underweight (p = 0.010) and had lower BMI values (p = 0.001). Compared with the patients with osteopenia, the patients with osteoporosis were older (p < 0.001), were more likely to have a history of previous bone fractures (p = 0.031) and increased PTH values.

Concerning nutritional parameters, autoimmune comorbidities (autoimmune thyroiditis, type I diabetes and dermatological diseases were the most frequent) and the presence of Marsh 3C at duodenal histology, no differences were found among the three groups.

A total of 16 females (7.5%) were in post-menopause at diagnosis, and a significantly greater proportion of the osteoporosis group were in post-menopause compared with the osteopenia (p < 0.001) and normal BMD groups (p < 0.001), respectively. Amenorrhea at diagnosis was present in three (1.4%) women, one in each of the three groups.

Seven patients chronically used drugs associated with the development of BMD alterations. Among these patients, three took corticosteroids for six to 10 months, while four had chronic use of SSRIs for two to eight years before they were diagnosed with CD. No differences in this variable were found among the three groups.

As shown in Table 2, compared with patients with normal BMD, osteoporosis was significantly associated with age ≥45 years (OR 6.5; 95% CI 1.3 to 32.2), male gender (OR 4.7; 95% CI 1.1 to 20.8), being underweight (OR 7.4; 95% CI 1.3 to 42.5) and having greater histological damage of the duodenum at diagnosis (OR 5.8; 95% CI 1.4 to 24.1). GI symptoms at diagnosis, menopause, alcohol intake and previous bone fractures were not significantly associated. Concerning increased PTH levels, the findings were just below significance, with an OR of 3.9 (CI 95% 0.97 to 15.5). The same logistic regression model was applied for osteopenia; compared with patients with normal BMD, the sole feature that was significantly associated with osteopenia was the presence of GI symptoms at diagnosis (OR 2.4; 95% CI 1.1 to 5.3; data not shown).

Table 2.

Risk factors for osteoporosis in adult coeliac disease (CD) patients in multivariate logistic regression analysis.

	Odds ratio	95% CI	p
Age
<45 years	1.0
≥45 years	6.5	1.3 to 32.2	0.022
Gender
Female	1.0
Male	4.7	1.1 to 20.8	0.039
Underweight (BMI <18.5 kg/m2)
No	1.0
Yes	7.4	1.3 to 42.5	0.025
GI symptoms leading to CD diagnosis
No	1.0
Yes	2.1	0.5 to 8.0	0.288
Higher histological damage (Marsh 3C)
No	1.0
Yes	5.8	1.4 to 24.1	0.014
Menopause
No	1.0
Yes	1.9	0.2 to 19.1	0.582
Previous fractures
No	1.0
Yes	4.1	0.9 to 18.7	0.069
Elevated PTH values
No	1.0
Yes	3.9	0.97 to 15.5	0.055
Alcohol (>3 units daily)
No	1.0
Yes	3.4	0.2 to 64.1	0.409

BMI: body mass index; CD: coeliac disease; CI: confidence interval; GI: gastrointestinal; PTH: parathormone.

Significant predictors are in italics.

Subanalysis of patients with antibodies positivity

Our CD population showed a percentage of CD-specific IgA antibodies negativity of 13.1% (n = 28 patients, both TTG IgA and EMA IgA negative). Among them n = 7 patients (3.27% of the total) presented with IgA-selective deficiency and were tested for other immunoglobulin classes (anti-TTG IgG, anti-DPG IgG), showing a positivity for at least one of them. To evaluate the possible influence of CD seronegative patients in the final results, a further subanalysis was carried out removing patients with seronegative CD without IgA deficiency (n = 21).

As showed in Table 1S, 38.7%, 43.6% and 17.7% of patients had normal BMD, osteopenia and osteoporosis, respectively. Features statistically different among the three groups were almost the same as in the overall population (Table 1).

At the logistic regression, performed using the same variables analysed in the overall population, osteoporosis was significantly associated with age >45 years (OR 7.8; 95% CI 1.2 to 51.6), male gender (OR 7.9; 95% CI 1.6 to 36.4), being underweight (OR 8.2; 95% CI 1.1 to 64.9) and having greater histological damage of the duodenum at diagnosis (OR 4.9; 95% CI 1.1 to 23.3) (Table 2S).

Male CD patient subgroup

To further investigate the male CD population, their characteristics were analysed. The male patients were divided into three groups according to the BMD score, and their features were compared (Table 3S). A total of 24.6% (n = 15) of men showed normal BMD while 45.9% (n = 28) and 29.5% (n = 18) presented with osteopenia and osteoporosis, respectively. Compared with men with normal BMD, men with osteoporosis had a higher median age (50.5 vs 36 years; p = 0.02) and were more frequently underweight (29.4% vs 0%; p = 0.04). In addition, they presented more frequently with anaemia (29.4% vs 0%; p = 0.04) and had lower values of ferritin (10.5 vs 88 ng/ml; p = 0.001) compared with the patients with normal BMD. Anaemia was also statistically more prevalent in the men with osteopenia compared with the men with normal BMD (33.3% vs 0%; p = 0.01). Compared with the patients with osteopenia, the men with osteoporosis more frequently had increased PTH values (58.8% vs 25%; p = 0.05) and lower values of ferritin (10.5 vs 44.5 ng/ml; p = 0.03). No differences were found between groups concerning GI symptoms leading to CD diagnosis, presence of symptoms >5 years before CD diagnosis, family history of CD, autoimmune comorbidities, previous bone fractures, drugs, smoking, alcohol (>3 AU daily) and Marsh 3C at diagnosis time. Regarding serological assays, no differences were found among the three groups of patients for positivity to anti-EMA and/or TTG autoantibodies, cholesterol, triglycerides and albumin values.

Discussion

This is a prospective study investigating a large series of adult CD patients (n = 214) who were actively screened for altered bone density at diagnosis and before the start of a GFD.

The observed prevalence of low BMD in this consecutively diagnosed cohort of adult CD patients was approximately 60%, in line with previously reported prevalence data (40%–75%).^4–10

The gender ratio reflected the usual distribution of CD (females:males = 2:1), making our cohort a suitable representation of the gender distribution in CD.^13,23 Previous studies considered primarily female coeliac patients (both pre- and post-menopausal), patients who had been on a GFD for variable periods of time or also included children in their evaluation, representing important biases in the assessment of the prevalence of BMD alterations in adults with CD.^12,15

In our cohort of newly diagnosed CD patients, males were more frequently affected by osteoporosis than females, with an OR of 4.7, highlighting male gender as a risk factor for osteoporosis in CD. The majority of papers studying the prevalence of osteopenia and/or osteoporosis in coeliac patients did not show a significant difference between genders.^24,25 Other studies that included a greater percentage of female CD patients, mostly post-menopausal, reported that women were more frequently affected by osteoporosis than men.^17,26 In two previous studies, a higher prevalence of osteoporosis in male CD patients than in pre-menopausal female CD patients was observed.^27,28 These data, together with our findings, suggest that male patients may benefit from investigations of altered BMD during the initial CD workup.

Underweight patients had a 7.4-fold higher probability of presenting with osteoporosis compared with normal weight patients. Our finding stresses the need for attention to this subgroup of patients and underlines the importance of BMI evaluation at the first clinical visit. The increased frequency of osteoporosis in underweight patients could be explained by more severe malabsorption that leads to deficiencies in micronutrients, including calcium and vitamin D.^29,30 Osteoporosis has been reported to be more common in patients with overt malabsorption.^12,15,26 In the present study, the presence of GI symptoms at diagnosis was associated only with osteopenia but not with osteoporosis; the values of major nutritional parameters, such as cholesterol, triglycerides, and albumin, were not significantly different in patients with normal vs altered BMD. These data show that, different from the findings for underweight, clinical presentation and routine nutritional parameters may not be able to detect the risk of osteoporosis.

Another factor associated with osteoporosis was greater histological duodenal damage (Marsh 3C) at diagnosis. This could be due to malabsorption secondary to the more severe intestinal mucosa damage and be associated with the chronic inflammation induced by CD.²⁹ Previous studies investigating the association between BMD and duodenal histopathology before the start of a GFD demonstrated that osteoporosis may develop even in cases of CD with mild histological damage (Marsh 1–2).³¹ Despite this evidence, osteoporosis was observed more frequently in coeliac patients with villous atrophy (Marsh 3 A–C) than in those patients with milder histological damage (Marsh 1–2).^32,33 Taking these findings together, we can hypothesise that intestinal damage may affect BMD, eventually leading to osteoporosis development.

Our data demonstrated that older age (≥45 years) at the time of CD diagnosis was a significant risk factor for osteoporosis. This finding, supported by previous studies,^5,8,9,34 is consistent with the known linear relationship between increasing age and the risk of low BMD.¹

Approximately one-third of the examined coeliac patients showed increased PTH values, with a statistically significant prevalence in the patients with osteoporosis in the univariate analysis. In the logistic regression, the results were just below significance, with an OR of 3.9 (CI 95% 0.97 to 15.5; p = 0.055), likely due to the number of lost patients included in the regression. This result shows that in newly diagnosed CD patients, evaluations of PTH values could be a useful part of the CD workup even if increased PTH does not automatically indicate osteoporosis.

Although post-menopausal women were significantly more represented in the group of patients with osteoporosis, menopausal status was not significantly associated with osteoporosis in the logistic regression. This might be related to the low prevalence of this condition in the studied population, which included many young patients. However, in our opinion, the lack of association between post-menopausal status and osteoporosis in this case series of CD patients does not affect the need for BMD investigation in this subset of patients independent of the presence of CD.

An increase in previous bone fractures was not significantly associated with osteoporosis in the logistic regression. An increased fracture risk (from 25% to 43%) in patients with CD compared with the general population has been described.^35,36 However, studies describing BMD values in CD patients with fractures and thus linking BMD values with the fracture rate are rare. Our result may be affected by the anamnestic collection of fracture history as patients might have remembered fractures without a well-documented recollection of the intensity of the trauma. In addition, it has been reported that BMD is not the only factor associated with fracture rates in coeliac patients.^12,14

Concerning the subanalysis of CD male patients, we can briefly comment that, in this subgroup of patients, statistically significant features of the osteoporotic group compared with the group with normal BMD overlapped to a great extent with those of the whole CD population (Table 3S).

An interesting finding in this subgroup of patients is the higher percentage of anaemia and the lower values of ferritin in male osteoporotic CD patients compared with those with normal BMD. Unlike female patients, in whom iron deficiency anaemia can be a confounding factor due to the possible concomitant presence of blood loss from altered menstrual flow, in male CD patients this alteration seems to be associated with osteoporosis.

This study has several strengths. First, the prospective design allowed the recruitment of a non-selected population of both genders; in addition, during the visit, detailed clinical data were collected for all patients, allowing comparisons of detailed clinical and anamnestic data among the groups of patients. A limitation of this study is that vitamin D and calcium assays were not available for comparisons among the three groups of patients.

In conclusion, BMD alterations were found in more than half (60%) of newly diagnosed adult CD patients. Osteoporosis was significantly associated with male gender, age over 45 years, being underweight and having greater duodenal histological damage (Marsh 3C), suggesting that in these patient subgroups, BMD evaluation at the time of CD diagnosis might be beneficial.

Declaration of conflicting interests

None declared.

Funding

This study was supported by a start-up research grant from University Sapienza of Rome 2017 (n. AR11715C5EC71A11).

Informed consent

Written informed consent was obtained from each patient.

Ethics approval

The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a prior approval by the institution’s human research committee. The study was approved by the local Ethics Committee Institutional Review Board, Sapienza University (9 November 2005).