| |
Fracture risk in hepatitis C virus infected persons:
Results from the DANVIR cohort study
|
| |
| |
Download the PDF here
Download the PDF here
" We conclude that HCV-exposed patients have substantially increased risk of all fracture types, while clearance of HCV only leads to a minor risk reduction ......In this large population-based, nationwide cohort study we observed a more than two fold increased risk of fracture in HCV-exposed patients compared to the general population and the risk was equally raised for low-energy and other fracture types. Furthermore the overall risk of fracture did not differ between patients with chronic vs. cleared HCV-infection...... Thus our study suggests that fracture risk is multi-factorial and indicate that a direct metabolic effect of HCV-infection on bone mineral density is not the major determinant of fracture risk among HCV-infected individuals. ......Established in 1968, the CRS maintains information on vital status and migration for all Danish residents [14]. We used CRS to select the comparison cohort and to obtain information on dates of death"
"We conclude that HCV-exposed patients have substantially increased risk of all fracture types, while clearance of HCV only leads to a minor risk reduction. Thus our study suggests that fracture risk is multi-factorial and indicate that a direct metabolic effect of HCV-infection on bone mineral density is not the major determinant of fracture risk among HCV-infected individuals. As the risk of low-energy and other fracture types were equally increased in the HCV-exposed population and the risk estimates were reduced when adjusting for alcohol and drug abuse, we suggest that lifestyle related factors have a substantial impact on the increased risk of fracture in the HCV-infected population. The main preventive measures to decrease risk of fractures in the HCV population should therefore focus on modification of life-style related factors."
"Compared with population controls, HCV-exposed patients had increased overall risk of fracture [adjusted incidence rate ratio (aIRR) 2.15, 95% Confidence Interval (CI) 2.03-2.28], increased risk of low-energy fracture (aIRR 2.13, 95% CI: 1.93-2.35) and of other fracture (aIRR 2.18, 95% CI: 2.02-2.34). Compared with cleared HCV-infection, chronic HCV-infection was not associated with increased risk of fracture at any site (aIRR 1.08, 95% CI: 0.97-1.20), or other fracture (aIRR 1.04, 95% CI: 0.91-1.19). The aIRR for low-energy fracture was 1.20 (95% CI: 0.99-1.44)."
"We used a cohort design to conduct two sub-studies: In sub-study 1, fracture risk was compared between Danish HCV-exposed patients and a population-based comparison cohort matched on age and sex. In sub-study 2, we used a cohort of HCV-antibody positive patients to compare fracture risk between HCV RNA-positive (viremic) and HCV RNA-negative (non-viremic) patients. Patients were included in the period 1995-2007 and followed until 2010......Sub-study 2 included all patients from the DANVIR cohort with a positive HCV antibody test and at least one test for HCV RNA after the positive HCV antibody test. Study inclusion was defined as the date of HCV RNA measurement. Rates of fractures were compared between HCV RNA-positive and HCV RNA-negative individuals. HCV RNA category was not changed to accord with the results of any subsequent testing"
Fracture risk in hepatitis C virus infected persons: Results from the DANVIR cohort study
Journal of Hepatology
July 2014
Ann-Brit Eg Hansen1,, Lars Haukali Omland2,, Henrik Krarup3, Niels Obel2,
on behalf of the DANVIR cohort study
1Department of Pulmonary and Infectious Diseases, Copenhagen University Hospital, Nordsjaellands Hospital, Hillerod, Denmark; 2Department of
Infectious Diseases, Copenhagen University Hospital, Rigshospitalet, Copenhagen, Denmark; 3Section of Molecular Diagnostics, Clinical
Biochemistry, Aalborg University Hospital, Aalborg Hospital, Aalborg, Denmark
Background & Aims
The association between Hepatitis C virus (HCV)-infection and fracture risk is not well characterized. We compared fracture risk between HCV-seropositive (HCV-exposed) patients and the general population and between patients with cleared and chronic HCV-infection.
Methods
Outcome measures were time to first fracture at any site, time to first low-energy and first non-low-energy (other) fracture in 12,013 HCV-exposed patients from the DANVIR cohort compared with a general population control cohort (n = 60,065) matched by sex and age. Within DANVIR, 4500 patients with chronic HCV-infection and 2656 patients with cleared HCV-infection were studied.
Results
Compared with population controls, HCV-exposed patients had increased overall risk of fracture [adjusted incidence rate ratio (aIRR) 2.15, 95% Confidence Interval (CI) 2.03-2.28], increased risk of low-energy fracture (aIRR 2.13, 95% CI: 1.93-2.35) and of other fracture (aIRR 2.18, 95% CI: 2.02-2.34). Compared with cleared HCV-infection, chronic HCV-infection was not associated with increased risk of fracture at any site (aIRR 1.08, 95% CI: 0.97-1.20), or other fracture (aIRR 1.04, 95% CI: 0.91-1.19). The aIRR for low-energy fracture was 1.20 (95% CI: 0.99-1.44).
Conclusions
HCV-exposed patients had increased risk of all fracture types. In contrast, overall risk of fracture did not differ between patients with chronic vs. cleared HCV-infection, although chronic HCV-infection might be associated with a small excess risk of low-energy fractures. Our study suggests that fracture risk in HCV-infected patients is multi-factorial and mainly determined by lifestyle-related factors associated with HCV-exposure.
Abbreviations:
HCV (hepatitis C virus), CI (confidence interval), aIRR (adjusted incidence rate ratio), BMD (bone mineral density), CRS (Civil Registration System), DNPR (Danish National Hospital Registry), ICD-8 (International Classification of Diseases, revision 8), ICD-10 (International Classification of Diseases, revision 8), IDU (intravenous drug use), CCI (Charlson Comorbidity Index), IRR (incidence rate ratio), PY (person years of observation)
Introduction
It is well known that advanced liver disease caused by hepatitis C virus (HCV) is associated with reduced bone mineral density (BMD) [1]. It remains controversial whether chronic HCV-infection has detrimental effect on bone health in the absence of advanced liver disease or other risk factors. Some studies have reported low BMD among non-cirrhotic HCV-infected patients [[2], [3]] while other studies have indicated that HCV-infected patients without significant liver disease or alcohol/drug use have BMD values similar to age-matched controls [[4], [5]].
Osteoporotic fractures have substantial public health impact [6] and it is therefore important to establish whether the reported increased risk of low BMD in HCV-infected individuals translate into increased incidence of fractures.
Only few studies have examined fracture risk in HCV-mono-infected patients. In a large cohort study, Lo Re et al. [7] observed increased risk of hip fractures among patients with HCV-infection compared with HCV-uninfected persons. This study did not report results for viremic vs. non-viremic patients and the mechanisms for the increased fracture risk are not well understood. It remains important to further examine the association between fracture risk in HCV-seropositive individuals and the contribution of chronic viremia vs. the contribution of lifestyle related risk factors associated with HCV-exposure such as illicit drug use, alcohol abuse, poor nutrition and increased risk of trauma.
In HIV-infection, it is well established that HCV-co-infection is associated with increased risk of fracture [[8], [9], [10]]. Recently, we extended these findings by demonstrating that HIV/HCV-co-infected patients had increased risk of both low-energy and high-energy fractures compared to both HIV-mono-infected patients and population controls [8]. The increased risk of high-energy fractures suggests that increased risk of trauma contributes to fracture risk in patients with HIV/HCV-co-infection. However HIV/HCV-co-infected patients represent an extremely marginalized group and the results may not be generalizable to HCV-mono-infected persons.
Additional data therefore are needed to clarify the association between HCV-infection and risk of fracture. In the present study, we compare the risks of fracture at any site, low-energy fractures, and other fractures including high-energy fractures in a Danish cohort of HCV-seropositive (HCV-exposed) patients and a population-based comparison cohort matched on age and sex. We further examine the incidence of low-energy and other fractures within the HCV cohort, comparing patients who were viremic (HCV RNA-positive) with those who were non-viremic (HCV RNA-negative) following a positive test for HCV antibodies.
Discussion
In this large population-based, nationwide cohort study we observed a more than two fold increased risk of fracture in HCV-exposed patients compared to the general population and the risk was equally raised for low-energy and other fracture types. Furthermore the overall risk of fracture did not differ between patients with chronic vs. cleared HCV-infection.
Both impaired bone strength and increased risk of traumas may contribute to the increased risk of fracture among HCV-exposed patients. The equally increased risk of all fracture types including non-low-energy fractures suggests that increased risk of trauma contributes substantially to the increased fracture risk. HCV-infection is a marker of past or on-going intravenous drug use and risk-taking behaviour [[26], [27]], and alcohol overuse and HCV-infection often coexist [28]. Thus a number of factors may increase risk of falls, violence, and traffic accidents in the HCV population [29].
Low BMD may be an effect of HCV-infection induced through inflammation or HCV induced liver disease [30]. Patients with cirrhosis, including HCV related cirrhosis, have a high prevalence of osteoporosis [[1], [31]]. In our study, adjusting for liver disease did not change the risk estimates substantially indicating that advanced liver disease is not a major driver of fracture risk in our populations. Low BMD may also be induced by risk factors associated with being exposed to HCV-infection including a lifestyle associated with malnutrition [32] and negative effects on bone metabolism associated with alcohol consumption [[33], [34], [35]]. Adjusting for HIV, IDU, alcohol abuse, and comorbidity diminished the fracture risk associated with being HCV-seropositive which implies that part of fracture risk can be attributed to these risk factors.
The finding of an increased risk of fracture in HCV-exposed patients was extended by comparing risk of fracture between patients with chronic and cleared HCV-infection. Of interest, fracture risk did not differ significantly between patients with cleared vs. chronic HCV-infection in adjusted analyses.
Accordingly, factors associated with HCV-infection rather than a direct effect of HCV-infection on bone health, seem to be main determinants of fracture risk. Albeit not a significant finding in adjusted analyses, viremic patients had approximately 20% higher risk of low-energy fracture than non-viremic patients, thus our study cannot discard that chronic HCV-infection per se also exerts negative effects on bone metabolism and thereby fracture risk, especially, when taking into account the rather few patients at the end of follow-up.
Recently, another large cohort study examined fracture risk in HCV-infected patients; this study found that HCV-infected persons had increased risk of hip fracture with a hazard ratio of 2.69 (adjusted hazard ratio of 1.47) compared with uninfected individuals [7]. The risk estimates are comparable with our results as we found a 2.96 fold increased risk of all low-energy fractures in HCV-exposed patients (adjusted IRR of 2.13). Lo Re and co-authors only reported data for hip fractures, and they did not report result according to HCV RNA status. Only two small studies in selected groups of post-menopausal women have previously compared fracture rates in patients with chronic vs. cleared HCV-infection. In contrast to our results, these studies found significantly increased fracture risk in HCV RNA-positive women compared with HCV RNA-negative women. The results may have been hampered by small number of endpoints [4] or may be confounded by shared risk factors for non-responding to interferon and being prone to fractures [36]. We did not specifically study postmenopausal women, but in stratified analyses, we found approximately the same relative risk estimates among persons older than 50 years or in women when comparing chronic vs. cleared HCV-infection.
To our knowledge, this is the first large-scale cohort study designed to compare fracture risk in viremic vs. non-viremic HCV-infected patients. The strengths of our study also include the population-based design with unique linkage to validated national registries, which allowed for long and complete follow-up.
Our study also had limitations. We estimated fracture risk starting from time of diagnosis of HCV-infection as we were not able to determine the date of infection. Regular testing for HCV RNA subsequent to an initial diagnosis is not performed systematically in Denmark. Therefore, we could not model HCV viremia as a time updated variable and our analyses did not account for spontaneous or treatment-related viral clearance nor reinfection during follow-up. However, spontaneous clearance of HCV-infection mainly happens shortly after the initial acute phase of the disease and occurs infrequently in later phases of the disease [37]. In addition, during the study period, only a minority of Danish HCV-patients received antiviral treatment [13]. We did not exclude patients with a diagnosis of fracture before study inclusion. Fracture data were incomplete before 1995 as the DNHR only included data on outpatient and emergency patient visits since 1995. The interpretation of our results should also take into considerations that fracture before the age of 50 years rarely reflects osteoporosis [6]. Another potential shortcoming is inaccuracies in fracture diagnoses reported to DNHR. However, the positive predictive value of the diagnoses registered in DNHR is generally high (70%-99%) [38], and has for hip fractures been shown to be as high as 93% [39]. Another limitation is the lack of data on known risk factors for osteoporosis, such as menopause, medication and smoking. If these factors were distributed unequally between study groups, which is likely to be the case in sub study 1, but less likely in sub study 2, they might have affected the results.
We conclude that HCV-exposed patients have substantially increased risk of all fracture types, while clearance of HCV only leads to a minor risk reduction. Thus our study suggests that fracture risk is multi-factorial and indicate that a direct metabolic effect of HCV-infection on bone mineral density is not the major determinant of fracture risk among HCV-infected individuals. As the risk of low-energy and other fracture types were equally increased in the HCV-exposed population and the risk estimates were reduced when adjusting for alcohol and drug abuse, we suggest that lifestyle related factors have a substantial impact on the increased risk of fracture in the HCV-infected population. The main preventive measures to decrease risk of fractures in the HCV population should therefore focus on modification of life-style related factors.
Patients and methods
Study design
We used a cohort design to conduct two sub-studies: In sub-study 1, fracture risk was compared between Danish HCV-exposed patients and a population-based comparison cohort matched on age and sex. In sub-study 2, we used a cohort of HCV-antibody positive patients to compare fracture risk between HCV RNA-positive (viremic) and HCV RNA-negative (non-viremic) patients. Patients were included in the period 1995-2007 and followed until 2010.
Setting
Denmark has a population of 5.4 million [11], with an estimated HCV prevalence of 0.4 % [12]. HCV-infected patients are treated in hospital departments that are specialised in infectious diseases, gastroenterology, or hepatology. Medical care, including antiviral treatment, is provided free-of-charge to all HCV-infected residents. Although with some uncertainty, it has been estimated that only 2% of the Danish HCV-infected population has been treated with interferon [13].
Data sources
We used the unique 10-digit civil registration number assigned to all Danes at birth or immigration [14] to link the following data sources:
Danish HCV cohort (DANVIR)
HCV-exposed patients were identified from the DANVIR cohort, which comprises all patients tested for HCV in 14 of the 18 laboratories that perform such testing in Denmark [[15], [16]]. DANVIR data includes results and dates of HCV antibody tests (from 1991 onward) and HCV RNA tests (from 1995 onward). While HCV antibody tests are performed in all participating DANVIR centres, most HCV RNA measurements are done in one centre (Department of Clinical Biochemistry, Aalborg University Hospital), as described previously [17]. When the DANVIR cohort was established, it was estimated to include more than 90% of all Danish patients tested for HCV RNA and the majority of patients tested for HCV antibodies [[15], [16]].
Civil Registration System (CRS)
Established in 1968, the CRS maintains information on vital status and migration for all Danish residents [14]. We used CRS to select the comparison cohort and to obtain information on dates of death.
Danish National Patient Registry (DNPR)
The DNPR, established in 1977, collects information on all non-psychiatric hospital admissions in Denmark. Data from outpatient and emergency department visits have been included since 1995. For each contact, the DNPR records dates of admission and discharge and up to 20 discharge diagnoses, assigned by physicians and coded according to the International Classification of Diseases, 8th revision (ICD-8) through 1993 and the 10th version (ICD-10) from 1994 onward [18]. We extracted data from the DNPR on fractures, comorbidity (including liver diseases), HIV-infection, alcohol abuse and drug abuse [15].
The Registry of Drug Abusers Undergoing Treatment has registered all individuals in Denmark assigned to treatment of drug addiction since 1996 [19]. Treatment of drug addiction in Denmark is restricted to referral centers, which must supply data to the registry in order to obtain funding.
Study population
We conducted two sub-studies to analyse the association between HCV-infection and risk of fractures.
· In sub-study 1 the study groups were the HCV cohort, which included all patients >16 years of age in the DANVIR cohort who after 1 January 1995 and after the age of 16 years tested positive for HCV (a positive test for HCV antibodies and/or HCV RNA), and the comparison cohort was extracted at random from CRS and consisted of 5 age- and sex-matched individuals who were alive and not registered with HCV diagnosis at study inclusion. For the HCV-exposed individual and the matched population controls study inclusion was the date of positive test for HCV in the index patient. Rates of fractures in the HCV and comparison cohorts were compared.
· Sub-study 2 included all patients from the DANVIR cohort with a positive HCV antibody test and at least one test for HCV RNA after the positive HCV antibody test. Study inclusion was defined as the date of HCV RNA measurement. Rates of fractures were compared between HCV RNA-positive and HCV RNA-negative individuals. HCV RNA category was not changed to accord with the results of any subsequent testing [[15], [20]].
Information on HCV-exposed patients and the comparison cohort
The following information on the HCV-exposed patients and persons in the comparison cohort was ascertained as of the date of study inclusion: comorbidity (± liver diseases), alcohol abuse, intravenous drug use (IDU), and HIV.
Information on alcohol abuse, IDU and HIV co-infection was obtained from DNPR and the Registry of Drug Abusers Undergoing Treatment (Please refer to Supplementary data for details). Comorbidity was included in the analyses as a modified Charlson Comorbidity Index (CCI) score derived from diagnoses registered in the Danish National Hospital Registry prior to the HCV diagnosis date [[21], [22]]. The CCI assigns a score between one and six to a range of diseases, with the sum of individual scores serving as a measure of patients' overall comorbidity. We identified comorbid diseases using the ICD-10 codes provided by Quan et al. [23] (matching ICD-8 codes to ICD-10 codes as closely as possible) (please see Supplementary Table 1 for details). Liver diseases were not included in the CCI in the present study. We defined three modified comorbidity levels based on the following CCI scores: none (score = 0), low (score = 1-2), and high (score ≥3).
Outcome
We had three outcomes: Time to fracture at any site, time to low-energy fractures and time to other fractures (please see Supplementary data for details). Low-energy fractures were defined as fractures possibly due to osteoporosis, typically those caused by low-energy trauma [[6], [8]].
Statistical analysis
Person-years at risk were computed from study inclusion until the first date of any fracture diagnosis, death, emigration or 1 January 2010, whichever came first. To assess the effects of HCV on the risk of fracture, we fitted cause-specific proportional hazards models for each of the three outcomes (any fracture, low-energy fracture, and other fracture) and computed hazard ratios with 95% confidence intervals (CI) as an estimate of incidence rate ratios (IRRs). In accordance with the matched design stratified models were used for sub-study 1. Individuals who died were censored on the date of death [24]. For each outcome two main models were used: model 1 in which no adjustments were made and model 2 including adjustment for HIV, IDU, alcohol abuse, and comorbidity. Further, we fitted two exploratory models, model 3 and 4 which were as model 2, but also adjusted for liver diseases (model 3) and previous fractures (model 4). In sub-study 2 models 2-4 were further adjusted for age and sex. The analyses were repeated in strata defined by sex, age (<50 years and ≥50 years at study inclusion) and absence of liver disease, and for sub-study 2, also presence of liver disease.
To estimate absolute risks of fractures we used the cumulative incidence function. In the analyses of any fracture, death was handled as competing risks, in the analyses of low-energy fracture, other fracture and death were handled as competing risks and in the analyses of other fractures, low-energy fracture and death were handled as competing risks [25]. SPSS, version 19 (IBM Inc), and R software, version 2.14.2, were used for data analysis.
Ethics
Our study was approved by the Danish Data Protection Agency.
Results
Descriptive data
Sub-study 1 included 12,013 HCV-exposed patients and 60,065 individuals from the comparison cohort with 72,356 and 454,115 person years of observation (PY) (Table 1). 64% were male and the median age at inclusion was 40 years (interquartile range: 31-48). HCV-exposed patients were more likely to have a diagnosis of alcohol abuse, IDU, HIV, other comorbidity, and a previous diagnosis of fracture than persons from the comparison cohort. Sub-study 2 included 7156 HCV antibody-positive patients with 42,390 PY, of whom 4500 (63%) were viremic and 2656 (37%) were non-viremic at the time of the HCV RNA test. Viremic patients were more likely to be male and to have a diagnosis of IDU and a previous diagnosis of fracture than non-viremic patients (Table 1).
Risk of fracture in HCV-exposed patients vs. the comparison cohort
HCV-exposed patients had an increased risk of fracture, low-energy fracture and other fractures compared to the comparison cohort (Fig. 1). Unadjusted IRRs (model 1) were 2.83 (95% CI: 2.70-2.97) for all fractures, 2.96 (95% CI: 2.73-3.22) for low-energy fracture, and 2.77 (95% CI: 2.61-2.93) for other fracture. Adjusted IRRs (model 2) were 2.15 (95% CI: 2.03-2.28) for all fractures, 2.13 (95% CI: 1.93-2.35) for low-energy fracture, and 2.18 (95% CI: 2.02-2.34) for other fracture. The estimates obtained in model 3 and 4 did not diverge substantially from that of model 2 (Supplementary Tables 2 and 3). As shown in Table 2, stratified analyses revealed that fracture risk was equally increased in the examined subgroups. The 10-year risk of low-energy fracture and other fractures for HCV-infected patients were 9.0% (95% CI: 8.4%-9.7%) and 17.4% (95% CI: 16.6%-18.3%) whereas the corresponding figures for the comparison cohort were 4.2% (95% CI: 4.0%-4.4%) and 7.9% (95% CI: 7.7%-8.2%) (Fig. 1).
Risk of fracture in viremic vs. non-viremic patients
In unadjusted analyses, HCV RNA-positive patients had a slightly increased risk of fracture compared to HCV RNA-negative patients: unadjusted IRRs were 1.18 (95% CI: 1.07-1.32) for all fractures, 1.28 (95% CI: 1.06-1.53) for low-energy fracture, and 1.14 (95% CI: 1.00-1.30) for other fracture. In adjusted analyses fracture risk did not differ significantly between the two groups: adjusted IRRs (model 2) were 1.08 (95% CI: 0.97-1.20) for all fractures, 1.20 (95% CI: 0.99-1.44) for low-energy fracture and 1.04 (95% CI: 0.91-1.19) for other fracture. The estimates obtained in model 3 and 4 did not diverge substantially from that of model 2 (Supplementary Tables 2 and 3). In the adjusted analyses we found no statistically significant increased fracture risk in any of the examined subgroups when comparing chronic vs. cleared infection (Table 3). The 10-year risk of low-energy fracture and other fractures for HCV RNA-positive patients were 9.5% (95% CI: 8.5%-10.6%) and 18.0% (95% CI: 16.6%-19.4%), whereas the corresponding figures for HCV RNA-negative patients were 7.7% (95% CI: 6.5%-9.0%) and 16.6% (95% CI: 14.9%-18.4%) (Fig. 2).
|
|
| |
| |
|
|
|
