Osteoporosis is a systemic skeletal disorder that is characterized by the loss of bone tissue, disruption of bone architecture, and bone fragility, leading to an increased risk of fractures. Age-related bone loss and estrogen deficiency after menopause are the main causes of osteoporosis. Bisphosphonates, which reduce fractures by suppressing bone resorption, are currently widely used in postmenopausal women with osteoporosis,1 and in breast cancer patients with bone metastases.2 Recently, data from preclinical studies have demonstrated that bisphosphonates, especially nitrogen-containing bisphosphonates, can reduce tumor burden in connective tissue and reduce cancer risk through affecting tumor apoptosis, proliferation, invasion, and angiogenesis.3–5 These findings implied that prolonged use of bisphosphonates might potentially reduce cancer development. However, the relationship between oral bisphosphonate use and the risk of developing cancer is still controversial. Recently published data suggest that the risk of esophageal cancer increases with 10 or more prescriptions for oral bisphosphonates and the use of oral bisphosphonate prescriptions over about a 5-year period. Similar results were also observed among individuals in Europe and North America.6 Furthermore, most studies focused on specific cancers such as breast cancer, colon cancer, esophageal cancer, and gastric cancer.7–10 Even though patients may take oral bisphosphonates for decades for treatment of osteoporosis, the evidence for the effect of long-term use of these drugs on cancer risk is limited. Moreover, few studies have focused on the association between oral bisphosphonates use and cancer risk in Asian populations. Whether taking oral bisphosphonates has general, rather than site-specific, antitumor effects remains undetermined. Alendronate, a nitrogen-binding bisphosphonate, is now commonly prescribed in postmenopausal women for treatment of osteoporosis worldwide, and is the most commonly prescribed medication for these patients in Taiwan.11 In the present study, we therefore set up a population-based, prospective case-cohort study to investigate the impact of using oral alendronate on the risk of developing new cancers in elder women with osteoporosis in Taiwan.
The National Health Insurance (NHI) program in Taiwan has been in operation since 1995 and enrolls nearly all the inhabitants of Taiwan. Currently, the National Health Insurance Research Database (NHIRD) at the National Health Research Institutes (NHRI) in Taiwan takes charge of the complete NHI claims database and has published several dozens of extracted datasets for researchers. The NHRI has released a cohort dataset made of 1,000,000 randomly sampled people who were alive during 2000 and collected all records on these individuals from 1995 onward. These random samples have been confirmed to be representative of the Taiwanese population. Each patient's original identification number has been encrypted to protect privacy. Because this cohort dataset consists of deidentified secondary data released to the public for research purposes, this study was exempt from full review by the Institutional Review Board. The encrypting procedure is consistent so that collection of several claims belonging to the same patient is feasible within the NHIRD. In Taiwan, bisphosphonate regimens including ibandronate and zoledronic acid have been approved for intravenous administration only and almost 95% of oral bisphosphonate users have been prescribed alendronate as the drug of choice for osteoporosis treatment.11
The study group of oral alendronate users entered into the NHIRD between January 1, 1998 and December 31, 2009 was composed of women who fulfilled the following criteria: (1) being over the age of 55 years; (2) having been diagnosed with osteoporosis (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 733.0 to 733.09); (3) having been exposed to oral alendronate for at least 1 month; and (4) not having exposure to oral steroids before enrollment. Patients diagnosed with malignancy (ICD-9-CM codes 140.0 to 208.91 and 230 to 239.9), vitamin D deficiency (ICD-9-CM codes 268–268.9), hyperparathyroidism (ICD-9-CM code 252.0), celiac disease (ICD-9-CM code 579.0), hyperthyroidism (ICD-9-CM code 242–242.91), and Paget's disease (ICD-9-CM code 731.0) before enrollment and using oral risedronate were excluded. Patients who received more than three prescriptions of intravenous bisphosphonates were also excluded. The date of first oral bisphosphonate prescription was the enrollment date. A control group was selected from those women with osteoporosis and age over 55 years without taking any kind of oral or intravenous bisphosphonates during 1996 to 2009 in the database. They were matched with the alendronate users of the study group in terms of age, sex, comorbidities, and the enrollment date by using stratified random sampling, which has been applied in our previous studies.12, 13
Bisphosphonates were identified by the Anatomic Therapeutic Chemical (ATC) code, which is an internationally accepted classification system for drugs coordinated by the World Health Organization Collaborating Center for Drug Statistics Methodology. Oral bisphosphonates approved to be prescribed for treatment of osteoporosis in Taiwan during 1996 to 2010 include alendronate (M05BA04) and risedronate (M05BA07). However, the number of risedronate users was small in the NHIR dataset because risedronate was only approved for prescription in April 2005. Most osteoporosis patients in Taiwan take alendronate,11 which was approved for prescription in October 1998. Our study focused on the comparison between alendronate users and matched controls. Data on the preparations prescribed, the date of prescription, and the numbers of packs/tablets prescribed were extracted and converted to defined daily doses (DDDs). The DDD system is a validated measure of drug consumption maintained by the World Health Organization.10 DDD is the assumed average maintenance dose per day of a drug used for its main indication in adults, which for oral bisphosphonates is the prevalence or treatment of osteoporosis. In order to further investigate the association between duration of oral alendronate use and cancer development, the total number of DDDs of oral alendronate received was categorized by the duration of exposure into ≤2 years (≤730 DDDs) and >2 years (≥731 DDDs) of alendronate use for subsequent analysis.
The recorded covariate variables included age, sex, preexisting hypertension (ICD-9-CM codes 401.0–405.99), diabetes mellitus (ICD-9-CM codes 250.0–250.93), chronic obstructive pulmonary disease (COPD) (ICD-9-CM codes 491–492.8 and 494–494.1), dyslipidemia (ICD-9-CM codes 272.0–272.4), chronic kidney disease (ICD-9-CM codes 581–587), coronary artery disease (ICD-9-CM codes 410–414.9), benign breast disease (ICD-9-CM codes 610–610.9), morbid obesity (ICD-9-CM code 278.01), aspirin use, and statin use. Similar identification of comorbidities has been applied in previous studies.12, 13
Cancer event measurement
The endpoint of the study was the first recorded administrative claim of cancer (ICD-9-CM codes 140.0–208.91 and 230–239.9), which was the main diagnosis during hospitalization or subsequent outpatient department visits. Diagnosis of cancer was mostly confirmed on the basis of the results of pathological findings for insurance purposes. The confirmation of cancers using insurance claims is valid and has been used in our previous study.14
Microsoft SQL Server 2005 (Redmond, WA, USA) was used for data management and computing. Statistical analysis was performed utilizing SPSS (Version 17.0; SPSS Inc., Chicago, IL, USA). All data were expressed as the frequency (percentage) for ordinal and categorical data and as mean ± standard deviation (SD) for continuous data. The continuous data between the study group and the comparison group were compared by Student's t test. The categorical data between them were compared with chi-square test and Yates' correction or Fisher's exact test as appropriate. Cancer-free survival analysis was assessed using Kaplan-Meier analysis, with the significance based on the log-rank test. Univariate regression analysis was carried out using Cox proportional hazards regression analysis. The Cox analyses were censored at death, the occurrence of event (cancer development), or the end follow-up among alive subjects without event. Statistical significance was inferred at a two-sided p value of <0.05.
Data were received from the NHIRD for 6906 women with osteoporosis and exposure to oral alendronate and 20,697 matched controls not using alendronate (Fig. 1). The mean age was 73.5 ± 8.4 years for alendronate users and 73.4 ± 8.4 years for controls. Both groups had a maximum follow-up period of 12 years. The mean follow-up duration was 4.3 ± 2.5 years and 4.9 ± 2.6 years in the study group and the control group, respectively. The underlying characteristics of the patients in these two groups were similar, as shown in Table 1. There was no significant difference in age and incidence of hypertension, diabetes, COPD, estrogen use, dyslipidemia, chronic kidney disease, colorectal polyp, benign breast disease, and morbid obesity between the two groups. Alendronate users are more likely to have coronary artery disease. Use of statins before the index date was higher in the alendronate users than in the controls.
Table 1. Baseline Characteristics of the Study Population
Chi-square test and t test and were used for categorical variables and continuous variables, respectively.
COPD = chronic obstructive pulmonary disease.
Age, years, mean ± SD
73.5 ± 8.4
73.4 ± 8.4
Hypertension, n (%)
Diabetes, n (%)
COPD, n (%)
Estrogen, n (%)
Dyslipidemia, n (%)
Chronic kidney disease, n (%)
Coronary artery disease, n (%)
Colorectal polyp, n (%)
Benign breast disease, n (%)
Morbid obesity, n (%)
Statin use, n (%)
Of the total sample of 27,603 patients, 3467 patients (12.6%) had newly-developed cancers during the mean 4.8-year period of follow-up. They included 821 patients from the study group (11.9%) and 2646 patients from the control group (12.8%). Table 2 shows the characteristics of patients with and without cancer occurrence during the follow-up period. The use of alendronate was insignificantly less in the cancer patients than that in cancer-free patients (23.7% versus 25.2%, p = 0.054). The association between oral alendronate and cancer development of different organ systems and specific origin are shown in Table 3. Oral alendronate users and nonusers had similar incidence of all malignancy, and specific cancers such as lung cancer, breast cancer, colorectal cancer, gastric cancer, esophageal cancer, bladder cancer, pancreas cancer, renal cell carcinoma, hepatoma, and cervical cancer. Subsequent Kaplan-Meier survival analysis with log-rank test showed that there was no there was no difference in the cumulative incidence of cancer-free survival between these two groups (p = 0.221) (Fig. 2).
Table 2. Characteristics of Subjects With or Without Newly-Developed Cancers During Follow-Up Period
After Cox proportional hazard model analysis adjusted for age, sex, and comorbidities, even in groups with different durations of exposure to alendronate, oral alendronate use was not associated with either increased or decreased risk of cancer incidence (Table 4). To further elucidate the relationship between oral alendronate use and future cancer development, different age groups, and treatment duration, as well as the concomitant use of estrogen were analyzed. In the different age groups, the association between the use of oral alendronate and cancer development remained insignificant (Table 5). The risk of cancer development due to the use of oral alendronate was further analyzed in patients who received and did not receive estrogen therapy for osteoporosis, and the risk remained similar in these two groups (Table 6). The hazard ratio of alendronate to cancer of different origin also demonstrated that oral alendronate use was not associated with an increased risk of cancer occurrence from different origin as well as total mortality (Fig. 3). Alendronate users had insignificantly higher risk of all-cause mortality compared to that of controls (adjusted hazard ratio, 1.07; 95% confidence interval [CI], 0.99–1.14; p = 0.075).
Table 4. All Malignancy Incidence in the Alendronate Users and Controls
The main finding of the current study is that there is no difference in the incidence of newly developed cancers between elderly women with osteoporosis exposed to oral alendronate and age- and comorbidity-matched unexposed controls.
Bisphosphonates inhibit osteoclast-mediated bone resorption and slow the release of calcium. They are mainly used to prevent or treat osteoporosis, particularly in postmenopausal women. Recent preclinical studies have shown that bisphosphonates had potent antitumor activity because they were able to reduce proliferation and induce apoptosis of tumor cell lines.15 They were also shown to inhibit tumor cell adhesion and invasion of the extracellular matrix in vitro. Wood and colleagues16 showed that zoledronic acid, a bisphosphonate with a heterocyclic imidazole substituent, has antiangiogenic properties and may be effective in the treatment of malignant bone disease. However, several clinical studies have demonstrated conflicting results on the effect of bisphosphonates on the occurrence of malignancies. Chlebowski and colleagues17 showed that oral bisphosphonate use was associated with significantly lower invasive breast cancer incidence in 2816 postmenopausal women. In their study, oral bisphosphonate use significantly increased the incidence of in situ breast cancer. In addition, oral bisphosphonates are known to irritate the lining of the esophagus, and similar esophageal irritation from the acid reflux has been linked to cancer, suggesting a possible association between oral bisphosphonates use and esophageal cancer. A study by Green and colleagues6 showed an increased risk of esophageal cancer for oral biphosphonates use, and the risk was significant higher for 10 or more prescriptions of oral bisphosphonates. The frequency of gastric cancer was not increased in their study. One possible explanation is that bisphosphonate users may have upper gastrointestinal side effects and more frequently undergo gastroenteroscopy examinations to find early esophageal cancers. However, subsequent studies demonstrated that the risk of being diagnosed with esophageal cancer was not increased, despite a higher frequency of upper endoscopies found in patients taking oral biphosphonates.10, 18 The discrepancy of malignancy of digestive organ also existed in gastric cancer. Cardwell and colleagues10 showed that there was no difference in risk of gastric cancer between bisphosphonate users and nonusers, but Abrahamsen and colleagues18 reported that alendronate user has a lower risk of gastric cancer occurrence as well as cancer mortality. It is difficult to elucidate clearly the discrepancy of cancer risk of esophagus and stomach in previous studies. Regarding colorectal cancer, Green and colleagues6 did not demonstrate this beneficial association with bisphosphonates, but recent published studies showed a reduced risk of colorectal cancer occurrence with prescription of bisphosphonates.8, 19, 20 It suggests that the relationship between oral bisphosphonate use and malignancy remains undetermined, and more evidence from different racial populations is needed.
To the best our knowledge, this is the first study to investigate the impact of chronic use of oral alendronate on cancer development in Asians. Our current study focused on elderly Asian women with osteoporosis. We only enrolled women with age over 55 years and assumed that most of the study subjects were in postmenopausal status.21 The association between cancer risk and postmenopausal women is still inconsistent. It has been reported that osteoporosis may increase the risk of colon cancer,22 and another study suggested a negative association between malignancy risk and menopause.23 Our study results show that there is no association between cancer development and use of oral alendronate, a widely used bisphosphonate in Taiwan. We show that there is no significantly increased risk of development of cancers, including lung cancer, bladder cancer, pancreas cancer, renal cell carcinoma, hepatoma, and cervical cancer, which had not been surveyed previously. This finding was in partial accordance with recent reports from the United States, Denmark, and the United Kingdom; these reports showed no significant increase in the risk of esophageal and gastric cancer in oral bisphosphonate users.10, 18, 24 In our study, the risk of cancer development with oral alendronate use was similar among users in different age groups and users of hormone replacement. This finding raises concerns regarding the application of oral bisphosphonate in variable individuals.
The main advantage of our study is the use of a population-based dataset, which enabled us to prospectively trace the differences between the study and control groups. Another advantage is the use of recorded prescription data rather than self-reported drug use, which may misclassify exposure. Because of either limited use volume of other bisphosphonate regimens in Taiwan, only oral alendronate, the most commonly prescribed medication for osteoporosis, was analyzed in our current study.11 Underestimation of alendronate usage is unlikely because of its relatively high cost and the inability to obtain it without prescription in Taiwan. The relationship between other bisphosphonate regimens and cancer risk should be clarified in subsequent studies. However, our study has some limitations. First, we could not obtain information on histological subtypes of the cancers, smoking, alcohol consumption, body mineral density levels, family history of cancer, and severity of osteoporosis. Those confounders were not available from the NHIR dataset and might bias the results. Second, we found that alendronate users have insignificantly higher risk of all-cause mortality in our study. In the NHIR dataset, we can get the information of cancer occurrence and withdrawals from database due to death. However, we did not know the incidence of cancer death and exact cause of death for each enrolled subject, especially subjects who died outside the hospitals. Therefore, it is difficult to investigate and explain the causal relationship between alendronate use and all-cause deaths. Osteoporosis and low bone mineral density are known to be associated with higher mortality.25 It is possible that alendronate users may have more severe osteoporosis which is associated with higher mortality risk as reported previously.25 Further larger prospective studies are needed to establish the causal relationship between oral alendronate and all-cause mortality. Third, the diagnosis of osteoporosis and cancer was based on the administrative claims data reported by physicians or hospitals. To avoid inaccurate diagnoses, we only selected those postmenopausal women who were diagnosed with osteoporosis by at least one orthopedist. In addition, although the first impression of cancer development was based on clinical manifestations, most hospitals in Taiwan capable of diagnosing and treating cancers have pathologists and access to techniques for specimen biopsy for pathological confirmation of cancers. In addition, the study population mainly comprised Taiwanese of Chinese descent, and these results might not be applicable to other populations. Further studies may be designed to investigate whether other kinds of bisphosphonates, such as etidronate, risedronate, and ibandronate, have antitumor effects in women with postmenopausal osteoporosis. Finally, the power of this study is relatively limited because of the limited study size, similar cancer risk in both groups, and low incidence of rare cancers. Further larger prospective studies or meta-analysis of randomized control trials are suggested to further verify our findings.
In conclusion, in the Taiwan NHIRD patient population, we found no evidence of a substantially increased risk of cancer development in elderly women with osteoporosis using oral alendronate.
All authors state that they have no conflicts of interest.
This study was partially supported by two intramural grants from Taipei Veterans General Hospital (V99B1-023 and V99B1-011) and from the National Science Council (A098125).
Authors' roles: All authors participated in the conception, design, interpretation of data, and drafting and revision of the manuscript.