Human Papillomavirus Infections and Increased Risk of Incident Osteoporosis: A Nationwide Population-Based Cohort Study

Patients with viral infections are susceptible to osteoporosis. This cohort study investigated the correlation between human papillomavirus (HPV) infections and the risk of osteoporosis via 12,936 patients with new-onset HPV infections and propensity score-matched non-HPV controls enrolled in Taiwan. The primary endpoint was incident osteoporosis following HPV infections. Cox proportional hazards regression analysis and the Kaplan-Meier method was used to determine the effect of HPV infections on the risk of osteoporosis. Patients with HPV infections presented with a significantly high risk of osteoporosis (adjusted hazard ratio, aHR = 1.32, 95% CI = 1.06–1.65) after adjusting for sex, age, comorbidities and co-medications. Subgroup analysis provided that populations at risk of HPV-associated osteoporosis were females (aHR = 1.33; 95% CI = 1.04–1.71), those aged between 60 and 80 years (aHR = 1.45, 95% CI = 1.01–2.08 for patients aged 60–70; aHR = 1.51; 95% CI = 1.07–2.12 for patients aged 70–80), and patients with long-term use of glucocorticoids (aHR = 2.17; 95% CI = 1.11–4.22). HPV-infected patients who did not receive treatments for HPV infections were at a greater risk (aHR = 1.40; 95% CI = 1.09–1.80) of osteoporosis, while the risk of osteoporosis in those who received treatments for HPV infections did not reach statistical significance (aHR = 1.14; 95% CI = 0.78–1.66). Patients with HPV infections presented with a high risk of subsequent osteoporosis. Treatments for HPV infections attenuated the risk of HPV-associated osteoporosis.


Introduction
Osteoporosis affects more than 200 million people and causes 8.9 million fractures annually [1], mainly in postmenopausal women [2] and in elderly man [3]. Subsequent population. All diagnostic records were validated by the Bureau of National Health Insurance (NHI) to ensure the accuracy of these data [25]. In this study, data were retrieved from the Longitudinal Health Insurance Research Database (LHIRD), a subset of NHIRD, which was composed of claims data of one million people randomly sampled from the study period between 1997 and 2013. The Institutional Review Board of Taichung Veterans General Hospital approved this study.

Study Population
In the present study, patients who had been clinically diagnosed with HPV infections between 2000 and 2013 were identified using International Classification of Disease Clinical Modification codes (ICD-9-CM code 079.4, 078.1, 078.10-078. 12 [26,27]. Only patients with diagnosis of HPV from at least one inpatient admission or three outpatient visits were selected. The index date was defined as the date of HPV infection diagnosis. Individuals with osteoporosis before the index date or within 1 year after the index date were excluded. Patients aged under 50 years old or beyond 100 years old, or patients with missing information on gender were excluded. The control group was selected from LHIRD, propensity score-matched in a 1:4 ratio by age, sex, index year, co-morbidities and co-medications, using the same protocol as described in previous studies [28][29][30][31]. In the subgroup analysis, patients with HPV infections who had received treatment procedures within three months after the index date were categorized as "with treatment" group, and otherwise as "without treatment" group. Such treatment procedures included: (1) electrocauterization for condyloma, (2) condyloma, excision, and electrocauterization, (3) CO 2 laser operation, (4) chemosurgery for condyloma, (5) simple or complicated electro cauterization, (6) liquid nitrogen cryosurgery, (7) simple or complicated cryotherapy that involved CO 2 freezing and liquid nitrogen. All participants were tracked until presence of osteoporosis, missing, death, or the end of the study, December 2013. A total of 12,936 subjects were included in the HPV group and 51,744 subjects were included in the non-HPV control group.

Primary Endpoint and Covariates
All subjects in both groups were tracked from the index date to the first osteoporosis event. The primary endpoint of this study was the occurrence of new-onset osteoporosis, which was diagnosed based on clinical history or spinal or hip bone mineral density (BMD) by dual-energy X-ray absorptiometry (DEXA) evaluation and the use of one of the following medications: denosumab, alendronate, risedronate, ibandronate, zoledronate, raloxifene, bazedoxifene, teriparatide, strontium ranelate [32]. Patients diagnosed with osteoporosis or the use of aforementioned medications before the index date or within one year after the index date were excluded. The use of osteoporosis medications as alternative representatives of osteoporosis patients, in addition to BMD or DEXA assessment, was aimed to intensify the diagnostic validity. To calculate the co-payment exemption of these medications, the NHI bureau used an auditing mechanism to minimize diagnostic uncertainty and misclassification of osteoporosis. To eliminate potential bias, factors including the demographic variables and relevant co-morbidities including hypertension, diabetes mellitus, hyperlipidemia, rheumatoid arthritis, chronic obstructive pulmonary disease (COPD), alcohol-related illness, chronic kidney disease (CKD), inflammatory bowel disease (IBD), HBV, HCV, cirrhosis, celiac disease, syphilis, HIV, chlamydia, gonococcus, hyperthyroidism, hyperparathyroidism, vitamin D deficiency, premature menopause, male hypogonadism, adrenal cortical steroids, smoking, and bone mineralization affecting [33] co-medications including long-term glucocorticoid use [34], phenobarbital, phenytoin, carbamazepine, heparin, warfarin, cyclosporine, tricyclic antidepressants (TCAs) or selective serotonin reuptake inhibitors (SSRIs), proton pump inhibitors, furosemide, thiazide, statin, and beta-blockers. Baseline characteristics within two years before the index date were retrieved and adjusted in the analyses.

Statistical Analysis
The chi-square test was used to compare the distribution of age, sex, index date, and baseline co-morbidities between the two groups, including those with versus without a previous history of HPV infections. The mean age of onset was compared using Student's t-test. The incidence rate of osteoporosis was estimated by dividing the amount of osteoporosis by follow-up person-years for HPV-infected cases and non-HPV controls. Multivariable Cox proportional hazards regression models were used to estimate the crude HRs (cHRs) and adjusted HRs (aHRs) for osteoporosis in patients with HPV infections, as compared to non-HPV controls. The covariates considered in the multivariable regression models included sex, age, index date, comorbidities, and co-medications. A multivariable Cox regression model adjusted for the covariates was used to reveal the effect of sex, age, and follow-up time on the incidence of osteoporosis. The HRs adjusted for covariates were calculated each subgroup, as stratified by sex, age, comorbidities, and co-medications. The cumulative incidence curve was obtained from the Kaplan-Meier method and examined by the log-rank test. All data analyses were performed using SAS (version 9.4; SAS Institute, Inc., Carey, NC, USA). The statistical significance level was set at p-value < 0.05 in the two-tailed test.

Basline Characteristics of Study Populations
In this retrospective cohort study, 12,936 HPV-infected patients and 51,744 propensity score-matched non-HPV controls were included. Baseline characteristics are shown in Table 1. The mean ages of the patients in the HPV and non-HPV groups were 63.26 and 63.13 years, respectively.

Discussion
In this 13-year nationwide population-based retrospective cohort study, there was a 32% increased risk of osteoporosis in patients with HPV infections, which was validated after adjusting for demographic variables, comorbidities, and co-medications. Patients who were women, aged between 60 and 80, of low and of high socioeconomic status, with long-term use of glucocorticoid, were susceptible to HPV-associated osteoporosis. Treatments for HPV infections lowered the risk of HPV-associated osteoporosis in patients with HPV infections.
Studies have shown that virus infections, such as HIV, HBV, HCV, and herpes zoster infections, were independently associated with a higher risk of osteoporosis [10,11]. Collectively, evidence from these studies provided the association between viral infections and the risk of osteoporosis. For instance, it was demonstrated that patients with herpes zoster infections presented with a 4.55-fold greater risk of osteoporosis [12], as associated with significantly high levels of interleukin (IL)-1b, IL-6, IL-8, IL-10, and tumor necrosis factoralpha (TNF-α) [39]. Among those inflammatory biomarkers, IL-6 was suggested as a potent stimulator of osteoclast-induced bone resorption and thus was central to the pathogenesis of bone loss in the context of chronic inflammation [40]. Likewise, HIV-infected individuals were reported to have lower BMD compared to non-infected controls [41], as supported by a meta-analysis that observed a significant increase in the risk of fractures in HIV-infected individuals with an incidence rate ratio of 1.58 (95% CI = 1.25-2.00) [41]. It was further demonstrated that a B-cell RANKL/osteo-protegerin-driven pathogenesis contributed to the compromised total hip and femoral neck BMD in ART-naïve HIV-infected patients, indicating that B-cell dysregulation promoted HIV-induced bone loss through an imbalance in the RANKL/osteo-protegerin ratio [42]. In the case of HBV infections, patients with HBV infections were shown to present with a significant great risk of osteoporosis [11], which was in accordance with observations that the seropositivity of the surface antigen for hepatitis B in adult men was significantly associated with lower BMD [43]. Chronic HBV infections can induce the production of inflammatory cytokines, such as TNF-α, IL-1, and IL-6, which increases RANKL that stimulates osteo-clastogenesis and bone resorption [44]. Moreover, TNF-α can inhibit osteoblast differentiation and promote osteoblast apoptosis [45], for which HBV infection-associated osteoporosis was proposed to be driven by inflammatory pathways that contributed to decreased bone formation, increased bone resorption, and a subsequently decreased systemic BMD [11].
The exact mechanism of how HPV infections undermined bone loss or osteoporosis has not been studied. However, highly expressed RANKL has been observed during the progression of HPV infection-associated cervical cancer, which was secreted by HPVinfected cells [22]. The excessive production of RANKL released by tumor cells was reported to trigger osteoclast-mediated bone destruction and to increase tumor burden [24]. Subsequently, binding of RANKL to RANK receptors on osteoclasts was found to activate signals for bone resorption [22]. On the other hand, high levels of inflammatory mediators such as TNF-α in patients with HPV infections [46] could place those individuals in an inflammatory microenvironment that may intensify osteoclastic resorption [47] by promoting RANKL production [48], transducing RANKL-induced signal pathways, and amplifying osteo-clastogenesis [47,49]. Specifically, TNF-α, as triggered by infections, promotes osteoblasts apoptosis and reduces osteo-blastogenesis by stimulating DKK-1 and Sost expression [9]. Moreover, TNF-α could suppress osteoblast differentiation by inhibiting Smad signaling through an NF-κB-mediated process [50]. Collectively, osteoclastogenesis in response to high concentrations of RANKL and TNF-α may explain bone resorption and osteoporosis in patients with HPV infections.
It was found in the present study that treatments for HPV infections attenuated the risk of HPV infections. This finding was in accordance with previous studies on benign lesions of anogenital warts [51], in which medical and surgical therapies were able to alleviate symptoms [52]. For instance, the primary clearance rate of lesions was estimated to be 44-87% for cryotherapy, 89-100% for scissor excision, 94-100% for electrocautery, and close to 100% for laser-assisted surgical treatments [53]. Moreover, as most HPV infections with genital warts could be eradicated within two years in immune-competent patients, early treatments of warts has been shown to exert higher clearance rates and lower incidence of malignancies [54], which may explain the observed beneficial effect of treatments for HPV infections on the reduced the risk of osteoporosis in the present study.
However, as treatments may not eradicate all HPV-infected cells, long-term follow-ups for patients with HPV infections would still be necessary. In particular, it was demonstrated that there was a recurrence rate of 20-30% after therapies on HPV-associated wart lesions, which could increase during follow-ups [53]. The recurrence rate of wart lesions after treatments for HPV infections was 12-42% at 1 to 3 months and 59% at 12 months following cryotherapy, 9-29% following scissor excision, 22% following electrocautery, and 17-19% at 3 months and 66% at 12 months following laser surgery [53]. All in all, early treatments for HPV infections and long-term follow-ups were recommended. Clinical implications also included screening [55] in patients with HPV infections and patient education [56][57][58][59][60][61].
The major strength of the present study was the use of longitudinal data of large sample size and a long follow-up duration that was able to provide the temporal association between HPV infections and the risk of new-onset osteoporosis, and the effect of treatments for HPV infections on the attenuated risk of HPV-associated osteoporosis. Findings in the present cohort study were representative of the general population, and potential measurable confounders were balanced through propensity score matching [58][59][60] for demographics, comorbidities, and co-medications. In addition, both the diagnoses of HPV infections and osteoporosis were adjudicated by physicians, which ascertained the accuracy of the diagnoses. The outcome measurement of osteoporosis was further validated via requiring the use of medications for osteoporosis as part of the criteria. In addition, subgroup analyses were used to elucidate effect measure modifications. That said, several limitations exist in this study. First, there could be underestimated cases of HPV infections if there were no clinically recognized lesions, which could have excluded patients with self-resolving or asymptomatic infections from the study population. Second, most of the Taiwanese population are East Asians, so findings of the present study may not be generalizable to other populations. Third, as information on serotypes of HPV, history of HPV vaccinations, and lifestyle determinants such as physical activity were not available in the dataset, further studies are warranted to address whether these factors could alter the risk of HPV-associated osteoporosis. Finally, the differentiation between HPV positivity alone and pre-cancerous lesions are not clarified. A stratified analysis based on the degree of lesion needs further execution.

Conclusions
In conclusion, patients with HPV infections were associated with a significantly greater risk of subsequent osteoporosis, especially in female patients, those aged between 60 and 80 years, and individuals with long-term use of glucocorticoids. Treatments for HPV attenuated the risk of HPV-associated osteoporosis. Prospective studies on the association between HPV genotypes and osteoporosis and studies of mechanistic approaches may shed light on HPV-associated bone loss and osteoporosis.  Informed Consent Statement: Patient consent was waived by the IRB (approval No. CMUH-104-REC2-115) because NHIRD was a de-identified database.
Data Availability Statement: Data were acquired from the NHIRD in Taiwan. The NHI database was available upon application.