High Prevalence of Osteopathy in Chronic Pancreatitis: A Cross-sectional Analysis from the PROCEED Study

Abstract

INTRODUCTION:

Chronic pancreatitis (CP) is associated with osteopathy (osteoporosis or osteopenia). However, existing literature is mostly limited to retrospective or administrative studies that have not clearly defined the prevalence and risk factors. Our aim was to identify patient- and disease-related associations with osteopathy in a prospective cohort study of CP.

METHODS:

We studied 282 subjects with definitive CP enrolled in the PROCEED study who had a baseline dual-energy X-ray absorptiometry (DXA) scan. Osteopenia and osteoporosis were defined using the lowest T-scores. Clinical data were collected using standardized case report forms. Comparisons were performed with a multivariate logistic regression model with forward selection to identify risk factors for osteopathy.

RESULTS:

The majority of subjects had osteopathy on DXA scan (56.0%; 17.0% osteoporosis; 39.0% osteopenia). Subjects with osteopathy had a higher prevalence of traumatic (40.0% vs. 26.4%, p=0.02) and spontaneous fractures (3.9% vs. 0, p=0.04). On multivariate analysis, older age (OR 1.29 per 5 years, 95% CI 1.15-1.45), female sex (OR 3.08, 95% CI 1.75-5.43), white race (OR 2.68, 95% CI 1.20-6.01), and underweight BMI category (OR 7.40, 95% CI 1.56-34.99) were associated with higher probability of osteopathy. There were no significant associations between osteopathy and patient and disease-related features of CP.

CONCLUSION:

In the largest study of CP patients who underwent DXA screening, the majority had osteopathy. There are overlapping risk factors with osteopathy in the general population, but the high prevalence in men and younger women supports the need for future investigations into the mechanisms of bone loss in CP. ClinicalTrials.gov number, NCT03099850.

INTRODUCTION

Chronic pancreatitis (CP) is a chronic inflammatory disorder that causes fibrosis and frequently endocrine and/or exocrine insufficiency¹. In this complex state of nutritional imbalance, patients can develop other systemic sequelae, including osteopenia and osteoporosis (collectively referred to as osteopathy)². Osteopathy is characterized by deterioration in the structural integrity of bone tissue and mass leading to bone fragility and increased risk of fractures. A previous meta-analysis including over 500 subjects reported a high pooled prevalence (65%) of either osteoporosis or osteopenia in subjects with CP³. Multiple studies have demonstrated an increased risk for bone fracture in CP compared to controls, emphasizing the need to better understand and prevent this common sequela of CP^4–6.

While the mechanism of reduced bone mineral density (BMD) in CP is not completely understood, a number of factors, including overlapping risk factors for both diseases, such as cigarette smoking and heavy alcohol use, and direct or indirect effects from CP may be responsible for this finding. CP may indirectly contribute to the risk of reduced BMD through exocrine dysfunction leading to micronutrient deficiencies (especially vitamins D and K) and low body mass index, as well as decreased physical activity. Unfortunately, the ability to assess for independent effects of these traditional risk factors on reduced BMD in CP has been hindered in previous studies by the inability to perform robust statistical comparisons (due to small sample sizes) or lack of access to patient level data (e.g., administrative databases and meta-analyses). Recently, Stigliano and colleagues reported analyses of subjects from seven European countries, and demonstrated increased odds of osteoporosis with increasing age, female sex, and decreasing body mass index, but these results are awaiting verification in an independent data set⁷. Lastly, there is a remaining need to assess the influence of pancreatic disease factors, such as duration of symptoms, CP disease severity, previous endoscopic or surgical procedures, and concurrent diabetes and/or exocrine pancreatic insufficiency.

PROCEED (Prospective Evaluation of Chronic Pancreatitis for Epidemiologic and Translational Studies) is a multicenter cohort study investigating the progression of acute and chronic pancreatitis⁸. All participants with definite CP are offered a DXA (dual-energy X-ray absorptiometry) scan at baseline, eliminating potential financial barriers to testing. This study provides a unique opportunity to analyze a large number of CP subjects with uniformly collected and detailed patient level data, including demographics, clinical characteristics of disease, and DXA imaging. In the current study, we aim to determine the baseline point prevalence and predictors of osteoporosis and osteopenia in CP.

METHODS

Study Design

The current study is a cross-sectional analysis of adult subjects (≥ 18 years old) with definite CP who were enrolled into an ongoing prospective cohort study, PROCEED (NCT03099850), between June 2017 and September 2020⁸. Subjects with available DXA scans at enrollment, including the preceding 3 years, were selected for study (Figure 1). One of the clinical outcomes of interest in PROCEED is osteoporosis, so all subjects with definite CP are recommended to have a DXA scan at baseline enrollment. If this is not completed as part of their routine clinical care, then it is offered at no cost as a research test.

Figure 1.

Participant flow chart.

Subjects were enrolled from nine clinical centers in the Consortium for the Study of Chronic Pancreatitis, Diabetes, and Pancreas Cancer (CPDPC) in the United States⁹. Demographic data, such as age at enrollment, race, and body mass index (BMI), as well as clinical characteristics were collected using a combination of self-administered forms, direct interview with the participant by a trained coordinator, and physician review of the medical record. These data (including imaging findings) were entered into an online database managed by a Central Data Management Center (CDMC; MD Anderson Cancer Center). The CDMC reconciled missing or discordant data with the study sites prior to performing statistical analyses. This study was approved by the Institutional Review Board of each of the study sites in the CPDPC.

Study Definitions

For the purposes of the PROCEED study, definite CP is defined by the presence of parenchymal or intraductal calcifications, a Cambridge classification of 3 or 4 on CT scan and/or MRCP imaging, or histology¹⁰. All radiologic studies (CT or MR) were reviewed by designated radiologists with subspecialty expertise in pancreatic imaging at each of the CPDPC clinical centers using a standardized data collection form and previously published reporting standards¹¹. Standard case report forms developed for PROCEED record patient and CP-disease specific data including age, body mass index (BMI), smoking status, history of alcohol use, CP duration, CP etiology, diabetes, pancreatic enzyme replacement therapy (PERT), previous endoscopic and surgical therapies, and history of either spontaneous or traumatic fractures. The etiology of CP (alcohol vs. other) was adjudicated based on the clinical impression of the treating physician. Diabetes status was based on the presence of a previous clinical diagnosis, the use of antidiabetic medications, and/or meeting American Diabetes Association criteria for abnormalities in fasting plasma glucose, HbA1c, and/or random blood glucose levels. Exocrine pancreatic dysfunction was assessed using a combination of a clinical history consistent with steatorrhea or results of fecal elastase testing (result of <100 mcg/gm of formed stool was considered as exocrine dysfunction). A fracture history was self-reported during adulthood, with fractures classified as either trauma (e.g., associated with minor, moderate, or major trauma) or spontaneous.

DXA Imaging

Participants enrolled into the PROCEED study with definitive CP were all requested to undergo a DXA if this had not been completed within 3 years of baseline enrollment. Scans were performed at the CPDPC clinical centers using standard clinical protocols, and data were recorded on a standardized case report form, including measurements of T-score at various anatomic locations (i.e., vertebra, hip, and/or radius). For the purposes of the current study, the assignment of normal bone mineral density (T-score >−1.0), osteopenia (T-score −1.0 to −2.5), or osteoporosis (T-score <−2.5) was based on the lowest T-score at any location. For this study osteoporosis and osteopenia were collectively referred to as osteopathy.

Statistical Analyses

Continuous data were summarized with means (SD) and compared using two-sample t-tests or Wilcoxon rank sum test, while categorical data were summarized using numbers (%) and compared with Chi-square test or Fisher’s exact test. The primary statistical outcome was the identification of osteopathy during the baseline period. The probability of osteopathy was modeled using univariate logistic regression with each of the patient- and disease-related covariates of interests. Then, covariates with a p-value <0.10 were entered into a multivariable logistic regression model. The final multivariate model was developed using forward selection to identify the most parsimonious model based on model fit using the c-statistic (area under the curve (AUC)). All tests were performed with SAS version 9.4 (SAS Institute, Cary, NC) and a 2-sided p<0.05 was considered statistically significant.

RESULTS

Study Cohort Characteristics

A total of 282 subjects with definitive CP enrolled in PROCEED who completed a DXA scan during the baseline study period were selected for further study. Study subjects had a median age of 56 years at enrollment (IQR, 47 - 64), a slight female preponderance (51.4%), and were primarily of white race (87.2%). Alcohol was the attributed etiology of CP in 41.6% of subjects, and 68.2% were either past or current smokers. Due to the nature of the study design, only a minority (8.9%) had a history of pancreatic surgery. Compared to CP subjects who did not have a baseline DXA scan, the study cohort was older at enrollment (median age 56 vs. 52, p<0.001), but there was a similar distribution of sex and mean BMI (data not shown).

Point Prevalence of Osteopathy and History Fractures

The majority of subjects (56.0%) had CP-related osteopathy, including 17.0% with osteoporosis and 39.0% with osteopenia. The prevalence of osteopathy exceeded 60% in subjects >50 years of age (Figure 2). The point prevalence of osteopathy is displayed in the CP study population according to age category (Figure 2A), sex distribution (Figure 2B), and age and sex combined (Figure 3) Importantly, prevalence rates remained >35% when the three youngest age categories were combined, and was close to 50% for men. A history of both traumatic (40.0% vs. 26.4%, p=0.02) and spontaneous fractures (3.9% vs. 0%, p=0.04) was more common in the osteopathy group.

Figure 2.

Point prevalence of osteopenia and osteoporosis in the CP study population according to age category (Figure 2A) and sex distribution (Figure 2B).

Figure 3.

Point prevalence of osteopathy (reflecting either osteopenia or osteoporosis) according to age and sex categorization in the CP study population. Numbers in x-axis labels reflect number of at risk subjects in each age category for both sexes.

Factors Associated with Osteopathy

On univariate analysis, increased age at enrollment, female sex, underweight BMI category, white race, past or present smoking, and PERT usage were associated with increased odds of having osteopathy (Table 1). Of these findings, the largest odds of osteopathy were the highest (OR 6.3) in underweight (BMI <18.5 kg/m²) CP subjects compared to those with normal weight (BMI 18.5 to <25 kg/m²); however, this represented a relatively small proportion of the study population. Other CP-related factors including an alcohol etiology, presence of diabetes, morphologic severity (presence vs. absence of calcifications), duration of CP, and history of endoscopic or surgical therapy were not associated with osteopathy (Table 1). On multivariate analysis, an underweight BMI category remained independently associated with a higher probability of osteopathy in subjects with CP, whereas being overweight or obese were associated with decreased risk (Table 2). Additionally, older age, female sex, and white race were independently associated with osteopathy.

Table 1.

Comparison of patient- and disease related factors in subjects with definitive chronic pancreatitis at the time of enrollment into PROCEED according to bone mineral density.

Variable	Osteopathy (n=158)	Normal (n = 124)	p-value
Demographics
Median age at enrollment (years)	59 (51, 65)	53 (44, 64)	<0.01
Female sex	95 (60.1%)	50 (40.3%)	<0.01
White race	144 (91.1%)	102 (82.3%)	0.03
Median BMI (kg/m2)	23.0 (20.4, 25.9)	26.6 (22.8, 29.7)	< 0.01
BMI category:
-low	23 (14.6%)	2 (1.6%)	< 0.01
-normal	87 (55.1%)	48 (38.7%)
-overweight	31 (19.6%)	44 (35.5%)
-obese	17 (10.8%)	30 (24.2%)
Clinical Characteristics
Smoking status:
-current	65 (41.1%)1	35 (28.2%)1	<0.01
-past	54 (34.18%)	37 (29.8%)
-never	38 (24.1%)	51 (41.1%)
History of acute pancreatitis present	121 (81.2%)9	95 (78.5%)3	0.58
Alcohol etiology of CP present	67 (42.7%)1	50 (40.3%)	0.69
Calcifications present on imaging	125 (79.1%)	91 (73.4%)	0.26
Atrophy of pancreas on imaging
- < 7 mm	22 (14.5%)6	19 (16.4%)8	0.73
- 7-14 mm	67 (44.1%)	46 (39.7%)
- >14-20 mm	50 (32.9%)	37 (31.9%)
- >20 mm	13 (8.6%)	14 (12.1%)
Duration of pancreatitis prior to enrollment (years)	5.0 (1.0, 10.0)5	5.0 (1.0, 11.0)2	0.92
Endoscopic therapy (prior or ongoing)	84 (53.9%)2	63 (51.2%)1	0.66
Previous pancreatic surgery	17 (10.9%)2	8 (6.5%)	0.20
Diabetes present	64 (41.8%)5	47 (39.2%)4	0.66
Exocrine pancreatic dysfunction present	84 (70.0%)38	59 (66.3%)35	0.57
PERT usage	100 (64.1%)2	61 (50.4%)3	0.02

Continuous variables are summarized as median (25^th, 75^th percentiles). Categorical variables are summarized as count (percentage of available data). The number of patients with missing data are annotated as superscript. The p-value is from comparison between normal and osteopathy groups, with Wilcoxon rank sum test for continuous variables and Fisher exact test for categorical variables. BMI: body mass index. CP: chronic pancreatitis. PERT: pancreatic enzyme replacement therapy.

Table 2.

Multivariate logistic regression models using patient- and disease-related factors to predict the probability of osteopathy among 282 subjects with chronic pancreatitis.

	Multivariate
Covariate	OR	95% CI	p-value
Age at enrollment (per 5 year increment)	1.29	1.15, 1.45	< 0.001
Female sex	3.08	1.75, 5.43	< 0.001
White race (vs. others)	2.68	1.20, 6.01	0.017
BMI groups (vs. normal weight):
Obese	0.18	0.08, 0.39	< 0.001
Overweight	0.32	0.17, 0.62	0.002
Underweight	7.40	1.56, 34.99	< 0.001

DISCUSSION

The current cross-sectional analysis from the CPDPC PROCEED study confirms the observation that osteopathy is common in CP. Specifically, osteoporosis or osteopenia were present in the majority (56%) of consecutive subjects who received a baseline DXA scan. Amongst a large number of risk factors assessed, increased age, female sex, white race, and lower BMI were independently associated with osteopathy. Notably, abnormalities were also observed in one-third of subjects younger than 50 at baseline. Although no CP-related clinical factors were associated with osteopathy, this apparent acceleration of bone loss, evident by an earlier chronological onset, suggests the need to further investigate the pathogenesis of osteopathy in CP.

Even though patients with CP are at an exceptionally high risk of developing decreased BMD, DXA screening is often not performed. For example, a recent study from an academic institution reported that only about 20% of patients with CP had completed a screening DXA scan¹². There are a number of factors that could potentially explain this discrepancy. First, DXA screening in CP is not consistently recommended by gastroenterological societies, in contrast to other diseases, such as inflammatory bowel disease and cholestatic liver diseases, which have comparable risks for decreased BMD. A consequence is that many clinicians remain unaware of the association of osteoporosis and CP. Although screening has been recommended by some groups, the lack of widespread societal endorsement also decreases the likelihood of insurance reimbursement, which may represent a financial deterrent for some patients^13–17. Lastly, considering the morbid nature of CP (e.g., its frequent association with moderate-severe abdominal pain), patients may prioritize other tests and appointments, and decline screening to assess bone health.

An association between CP and osteopathy was introduced more than two decades ago¹⁸. The relatively slow progress in characterizing the prevalence of this complication and its associated risk factors has been hindered due to the low disease prevalence of CP and limited availability of data from DXA testing. Even though this appears to be a common problem, occurring in approximately two-thirds of patients with CP, individual studies have generally involved small sample sizes, which limit the ability to perform robust statistical comparisons. Accordingly, it has become increasingly clear that epidemiological studies of CP are well suited for investigation through multi-center collaborations. For example, a team of investigators from 7 European countries recently published their pooled data (n=211), which demonstrated osteopathy (defined as either osteoporosis or osteopenia) in 64% subjects diagnosed with CP using the M-ANNHEIM criteria⁷. The prevalence of osteoporosis varied considerably between countries (range 0-31%) raising the possibility of variation due to ethnicity. The current study from the CPDPC investigators involves a somewhat larger sample size of subjects enrolled from a single country, and demonstrates remarkably similar results in regards to both prevalence rates and factors (age, sex, and BMI) associated with osteopathy, verifying these previous findings. Although subjects were predominantly white (87%) in the current study population, we were also able to identify that white race was associated with an increased risk for osteopathy; an observation that is consistent with previous reports of increased risk for fracture in other study populations^19–21. Further analyses in a more racially diverse study population of CP will improve the robustness of this observation as well as other complications associated with CP²².

The observation of a moderately high prevalence of osteopathy amongst younger patients and in men warrants further validation, because it may have important clinical implications. For example, the United States Preventive Services Task Force (USPSTF) recommends reserving DXA screening for postmenopausal women, women ≥65, or women <65 at high-risk (which does not include a diagnosis of CP)²³. In the current study population a diagnosis of osteopathy would have been missed in 46% (63 out of 137) men. Since the menstrual status for female subjects was unknown it is not possible to definitively calculate a miss rate; however, considering the average age of natural menopause is >50, the diagnosis would have been missed in 50% (23 out of 46) of women <50 who were most likely pre-menopausal. The observed prevalence of osteoporosis and osteopenia observed in our study in these age and sex categories is markedly higher than anticipated. For example, in a recent study from >15,000 subjects enrolled in the National Health and Nutrition Examination Survey (NHANES) the prevalence of osteoporosis (excluding osteopenia) was 5.6%, with much lower rates in the subgroups of subjects <60 years of age (1.7%) and men (2.5%)²⁴. Similarly, a study of premenopausal women (ages 20-44 years) reported osteoporosis and osteopenia in 0.2% and 13.1%, respectively²⁵. Missed diagnoses from not screening men and pre-menopausal women with CP represents a potential opportunity for earlier lifestyle and medical interventions to prevent osteoporotic fractures. Our current analysis of the rate of fractures is limited due to the cross sectional nature of this analysis; however, a prior observational study in a predominantly male Veteran’s Affairs (VA) population demonstrated the risk of hip fracture is three times higher in CP compared to non-CP controls emphasizing the importance of reducing the rates of missed or delayed diagnoses⁶.

While postmenopausal osteoporosis is readily attributable to increased bone remodeling due to estrogen deficiency, alternative explanations are needed for a sizeable portion of the current study population. Malnutrition in CP has been proposed as a potential contributing factor to osteopathy. We did not observe differences in subjects with a preceding clinical diagnosis of exocrine pancreatic insufficiency (EPI) or use of pancreatic enzyme replacement therapy (as an alternative surrogate for EPI), which may be a reflection of the sample size of these subgroups. On the other hand, a low (underweight) BMI was strongly associated with osteopathy, which was independent of the CP etiology (i.e., it was not due to confounding from an alcohol etiology). In CP a low BMI is often multifactorial, and more detailed studies are needed using simultaneous assessments of food security, dietary intake patterns, rate of absorption, and metabolic rate to understand the complex interaction of these and other factors. The current study design did not allow us to definitively assess the potential effects of vitamin D deficiency. There were observed trends toward more frequent use of vitamin D supplementation (30.9% vs. 21.7%, p=0.09) and higher mean vitamin D levels (35.2 ng/dL vs. 29.7, p=0.05) in the group with osteopathy. However, this likely reflects reverse causality related to institution of supplementation following recognition of osteopathy. Future studies with serial assessment of serum vitamin D levels prior to and at the time of DXA imaging may reduce this potential confounding.

There are preliminary data to suggest CP is associated with a state of increased bone turnover, which is hypothesized to be a consequence of chronic inflammation. One study demonstrated increased markers of bone resorption (serum C-telopeptide crosslink (CTX)) and formation (osteocalcin and procollagen 1 amino-terminal propeptide (P1NP)) in 29 cases with CP compared to 29 healthy controls²⁶. In the CP subjects with osteopathy there were also significant increases in interleukin-6 (IL-6), which has been consistently implicated in the pathogenesis of CP²⁷. Prior studies have demonstrated that IL-6, IL-1, and TNF-alpha may upregulate RANKL (receptor activator of nuclear factor-kappa B ligand) on osteoblasts, which directly leads to activation of osteoclasts and bone resorption^{28, 29}. The RANK-RANKL dimerization and osteoprotegerin system is a key regulator of osteoclastogenesis and is an important link between the immune system and bone metabolism. This system has been implicated as a primary mediator of the development of osteopathy in patients with inflammatory bowel disease and cirrhosis^{30, 31}. Future translational studies are needed to assess the contributions of IL-6 and other mediators of systemic inflammation in CP to examine alternative explanations for the apparent acceleration of bone loss that is not accounted for by clinical features of disease, such as the presence of calcifications or duration of CP.

There are a number of advantages to the current study, including the multi-center collaboration from nine NIH-supported clinical centers in the United States, which allowed us to enroll a large number of patients with CP to permit more comprehensive statistical analyses. The prospective collection of a large number of variables using standardized case report forms, along with central data management, assured high data quality. Selection bias related to studying patients at highest risk for osteopathy has been a concern with all prior studies in CP. The ability to offer DXA screening to participants as a protocol related test allowed us to reduce the risk of this bias and include a broader range of subjects, particularly those who would not routinely qualify for screening thereby increasing the generalizability of our results within a patient population of CP.

Despite these strengths, there are some limitations that must be considered to accurately interpret our results. First, despite concentrated efforts, slightly more than half of all CP subjects (55%) enrolled into PROCEED completed a baseline DXA scan and could be included in the current analyses. The study population with DXA scans had an older age at enrollment, which suggests the prevalence of osteopathy may potentially be overestimated. Conversely, the use of DXA screening in patients who otherwise may be at low risk (e.g., men and younger women) may negate this bias and partially explain the lower estimate of osteopathy compared to prior studies. While previous groups have examined the potential influence of fat soluble vitamin deficiencies (especially vitamins D and K), these data are not available for analysis in the current study, and remains an important area of focus for future studies^{7, 32}. Second, we were unable to confidently assess for the impact of exocrine pancreatic insufficiency in the study population, which is a common challenge in CP research investigations due to the absence of an accurate and convenient method of assessing for this complication³³. For this study we tried to minimize misclassification by assessing for the presence of exocrine pancreatic dysfunction using a composite of subjects with a clinical diagnosis (due to the presence of overt steatorrhea) and/or a low fecal elastase-1 level, independent of PERT usage. Lastly, the current cross-sectional study design does not permit determination of rates of bone loss or an accurate assessment of the lifetime risk for spontaneous fractures, which remains the ultimate clinical outcome of concern. Longitudinal follow-up is planned as the cohort matures to further understand the cumulative burden of osteopathy in CP.

In summary, in this cross sectional analysis from a large, multicenter study we demonstrated that two-thirds of subjects with CP have osteopathy (i.e., either osteoporosis or osteopenia). Advancing age, female sex, white race, and lower BMI were associated with increased odds of having osteopathy. However, the observation of clinically concerning rates of osteopathy in men and younger women may reflect an acceleration of bone loss in CP. Additional investigations are needed to further characterize and understand the mechanistic underpinnings of decreased bone mineral density in CP to identify opportunities for prevention and/or treatment. Similarly, longer term follow-up studies are awaited to more fully estimate the cumulative disease burden from osteopathy in CP.

Abbreviations:

BMI

body mass index

CDMC

Central Data Management Center

CP

chronic pancreatitis

CPDPC

Consortium for the Study of Chronic Pancreatitis, Diabetes, and Pancreatic Cancer

DXA

dual-energy X-ray absorptiometry

EPI

exocrine pancreatic insufficiency