Urinary tract infection increases subsequent urinary tract cancer risk: A population‐based cohort study

Abstract

This study examines the possible association between urinary tract infection (UTI) and urinary tract cancer (UTC). Data from the National Health Insurance system of Taiwan were used for the analysis. The UTI cohort included 70 116 patients who were diagnosed and recruited between 1997 and 2010. Each patient was randomly frequency‐matched with two people without UTI from the general population based on their age, sex, and month of UTI diagnosis. Cox's proportional hazard regression analysis was used to estimate the effects of UTI on UTC risk until the end of follow‐up on December 31, 2010. Patients with UTI had a significantly higher risk of developing UTC than healthy people (adjusted hazard ratio, 4.66; 95% confidence interval, 3.55–6.10). Further analyses indicated that risks are potentially related to the level of the lesion site. This study indicated that patients with UTI had a higher risk of developing UTC.

Urinary tract infection (UTI) is the most common extraintestinal infectious disease entity in women worldwide. Approximately 50% of women have at least one symptomatic UTI during their lifetime, and recurrent episodes are common.1 Several infections are related to the development of cancer, including human papilloma virus‐associated cancer such as head and neck cancers, cervical cancer,2, 3 hepatitis B and C virus‐associated liver cancer,4, 5 Epstein–Barr virus‐related nasopharyngeal carcinoma and lymphoma,6 and Helicobacter pylori infection‐induced gastric cancer.7

Chronic inflammation induced by bladder infection could have a role in human bladder cancer development.8, 9 The relationship between schistosomiasis and bladder cancer has been confirmed.10 Several case–control studies focused on the connection between UTI and bladder cancer risk and, although most showed that patients with UTI had a higher risk of bladder cancer, results were inconsistent.11, 12 Compared to bladder cancer, significantly fewer studies have discussed the connection between UTI and renal pelvis/ureter cancer because of its relative rarity. Studies regarding this issue showed either non‐significant findings13 or inconclusive results.14

Several mechanisms were proposed for explaining the association between infection and bladder cancer.11, 12 First, chronic inflammation of the bladder may induce urinary retention and stasis, which can potentially increase exposure and absorption of carcinogens present in the urine.15 Second, bacterial flora in the urine may contribute to the production of nitrites that are converted to carcinogenic nitrosamines.16 This study attempts to determine if there is a higher risk of bladder cancer for patients with UTI by examining a Taiwanese cohort. Because of an unusually high incidence of upper urinary tract urothelial carcinoma in Taiwan,17upper urinary tract cancer was of primary interest, but ensuing investigations were extended to cover all urinary tract cancers (UTC) except renal cell carcinoma. The results in this study were generated from a population‐based retrospective cohort obtained from the National Health Insurance (NHI) system's database in Taiwan.

Materials and Methods

Data sources

This longitudinal retrospective cohort study used reimbursement data obtained from the medical claims database of the Taiwanese universal NHI system. The NHI was implemented in 1995, providing health care to approximately 99% of the population (23.74 million people). The NHI system has contracts with 97% of hospitals and clinics in Taiwan.18 The claims data from the NHI Research Database (NHIRD), which is managed by the National Health Research Institute (NHRI), was used for this study. The NHRI assembled a research dataset containing all reimbursement claims records from 1996 to 2010 for one million randomly selected insured residents. Data from the NHIRD are representative of the entire population because the NHRI reported that no statistically significant differences were found in the distributions of age, sex, or health‐care expenditures between the subset of the NHIRD and all enrollees. Personal identification information was encrypted before releasing the research database to protect patient privacy. This study was approved by the Ethics Review Board of China Medical University (Taichung, Taiwan) (CMU‐REC‐101‐012).

Study patients

Patients with recently identified UTI (1997–2010) (International Classification of Diseases, 9th revision, Clinical Modification [ICD‐9‐CM] codes 590, 595) were selected as the study cohort from the database of one million randomly selected insured residents. A non‐UTI comparison group was randomly selected from all NHI beneficiaries aged 20 years and older and matched with the UTI group at a 2:1 ratio based on age (every 5‐year span), sex, month of UTI diagnosis, and area of residence. The index date was used as the date of UTI diagnosis. The 15th day of the index date month was used for the comparison group. Patients with any cancer (ICD‐9‐CM codes 140–208) before the index date or aged younger than 20 years were excluded from the study.

Outcome measures

The confirmation of UTC (ICD‐9‐CM codes 188, 189) events was based on the registry of catastrophic illness, which is a sub‐dataset of the NHIRD. Each patient was followed until a diagnosis of UTC was confirmed, until the patient was censored for loss at follow‐up, death, withdrawal from the insurance system, or to the end of the follow‐up (December 31, 2010). Some of the UTI may be caused by the UTC itself. To consider this possibility, we excluded the UTC cases diagnosed within 1 year after recruitment. Moreover, the baseline comorbidity history of each patient was identified, including diabetes (ICD‐9‐CM code 250), hypertension (ICD‐9‐CM codes 401–405), and hyperlipidemia (ICD‐9‐CM codes 272).

Statistical analysis

Data were analyzed using the sas statistical software package for Windows (version 9.1; SAS Institute, Cary, NC, USA), and the significance level was set at 0.05. Demographic factors such as age, sex, and comorbidities were compared between the UTI cohort and non‐UTI cohort. The differences were tested with the Chi‐squared test for categorical variables and Student's t‐test for continuous variables. The incidence of UTC was compared between the UTI cohort and the non‐UTI cohort stratified by sex, age, and comorbidities. Cox proportional hazard analysis was used to estimate the hazard ratio (HR) and 95% confidence interval (CI) of UTC in the UTI and non‐UTI cohorts. Multivariate Cox proportional hazard regression models were used to assess the UTC risk associated with demographic factors and comorbidities. We used the Kaplan–Meier method to estimate the failure functions of UTC between the UTI cohort and the non‐UTI cohort, and assessed the differences using a log–rank test.

Results

The characteristics of all 209 838 participants from the UTI cohort and the non‐UTI comparison cohort are listed in Table 1. Of all study participants, 78.64% were women, and most (61.95%) were aged 50 years and younger. Patients with UTI were more likely to reside in northern areas of Taiwan (41.96%). Patients in the UTI cohort had a higher incidence of diabetes (10.83% vs 6.43%; P < 0.0001), hypertension (23.64% vs 17.09%; P < 0.0001), and hyperlipidemia (13.36% vs. 8.07%; P < 0.0001), compared to the non‐UTI comparison cohort.

Table 1.

Comparison of demographics and comorbidity between patients with urinary tract infection (UTI) and non‐UTI controls

	Urinary tract infection				P‐value
	No (N = 139 722)		Yes (N = 70 116)
	n	%	n	%
Age, year
≤50	86 670	62.03	43 335	61.80	0.39
51–64	28 948	20.72	14 521	20.71	–
>65	24 104	17.33	12 260	17.50	–
Mean (SD)	45.83	17.19	46.39	16.92	<0.0001†
Follow‐up years, mean (SD)	8.94	4.01	9.23	3.82	<0.0001†
Sex
Female	109 842	78.61	55 169	78.77	0.72
Male	29 880	21.39	14 947	21.23	–
Area of residence
Northern	58 700	42.01	29 352	41.87	0.88
Medium	29 864	21.37	15 046	21.55	–
Southern	39 261	28.10	19 698	28.11	–
Eastern	11 897	8.51	6020	8.59	–
Comorbidity
Diabetes	8986	6.43	7594	10.80	<0.0001
Hypertension	23 879	17.09	16 572	23.64	<0.0001
Hyperlipidemia	11 279	8.07	9370	13.36	<0.0001

Chi‐square test; †t‐test.

Overall, the incidence of UTC was higher in the UTI cohort than in the non‐UTI cohort (2.89 vs 0.60 per 10 000 person‐years, HR = 4.80, 95% CI = 3.67–6.28). The multivariate Cox proportional hazard regression model revealed that patients with UTI had a 4.66‐fold higher risk of developing UTC (adjusted HR = 4.66, 95% CI = 3.55–6.10) after considering age, sex, and comorbidities. Both cohorts showed that men had a higher incidence of UTC than women (1.31 vs 0.43 per 10 000 person‐years, and 8.25 vs 1.62 per 10 000 person‐years for non‐UTI and UTI cohorts, respectively), and had a higher HR than women (6.30 with 95% CI = 4.24–9.37 vs HR = 3.78 with 95% CI = 2.62–5.45). The adjusted HR was significantly higher in men than in women (2.78, 95% CI = 2.17–3.56) (Table 2). The HR was highest in patients aged 65 years and more (HR = 5.63, 95% CI = 3.83–8.27). Thus, the adjusted HR was 11.8 (95% CI = 7.81–17.8) for elderly patients, compared to patients ≤50 years of age. Compared to patients living in northern areas, the patients lived in western, southern, and eastern areas had higher risks of developing UTC.

Table 2.

Incidence and adjusted hazard ratio of urinary tract cancer stratified by sex, age, and comorbidity compared between urinary tract infection (UTI) and non‐UTI

Variables	Urinary tract infection						Crude HR‡ (95% CI)	Adjusted HR§ (95% CI)
	No			Yes
	Case	PY	Rate †	Case	PY	Rate †
All	75	1 248 655	0.60	187	646 960	2.89	4.80 (3.67, 6.28)***	4.66 (3.55, 6.10)***
Sex
Female	43	1 003 982	0.43	85	523 254	1.62	3.78 (2.62, 5.45)***	1.00 (Reference)
Male	32	244 673	1.31	102	123 706	8.25	6.30 (4.24, 9.37)***	2.78 (2.17, 3.56)***
Age
≤50	14	805 109	0.17	20	421 827	0.47	2.70 (1.37, 5.35)**	1.00 (Reference)
51–64	26	265 329	0.98	66	133 666	4.94	5.04 (3.20, 7.94)***	6.25 (4.16, 9.40)***
>65	35	178 216	1.96	101	91 467	11.04	5.63 (3.83, 8.27)***	11.8 (7.81, 17.8)***
Area of residence
Northern	14	513 782	0.27	43	268 778	1.60	5.83 (3.19, 10.6)***	1.00 (Reference)
Medium	10	270 308	0.37	35	139 965	2.50	6.72 (3.33, 13.6)***	1.27 (0.86, 1.88)
Southern	43	355 317	1.21	79	182 122	4.34	3.59 (2.47, 5.20)***	2.55 (1.86, 3.49)***
Eastern	8	109 248	0.73	30	56 095	5.35	7.30 (3.35, 15.9)***	2.36 (1.56, 3.56)***
Comorbidity
Diabetes
No	62	1 181 417	0.52	149	586 855	2.54	4.82 (3.58, 6.48)***	1.00 (Reference)
Yes	13	67 237	1.93	38	60 105	6.32	3.20 (1.70, 6.00)***	1.13 (0.93, 1.66)
Hypertension
No	45	1 061 009	0.42	87	511 630	1.70	3.99 (2.78, 5.72)***	1.00 (Reference)
Yes	30	187 646	1.60	100	135 330	7.39	4.57 (3.04, 6.87)***	1.37 (1.04, 1.80)*
Hyperlipidemia
No	64	1 160 834	0.55	138	570 264	2.42	4.37 (3.25, 5.88)***	1.00 (Reference)
Yes	11	87 821	1.25	49	76 696	6.39	5.04 (2.62, 9.69)***	1.27 (0.86, 1.88)

*P < 0.05, **P < 0.01, ***P < 0.0001. †Incidence rate, per 10 000 person‐years; ‡related hazard ratio; §multivariable analysis including age, sex, and comorbidities. CI, confidence interval; HR, related hazard ratio; PY, person‐years.

All the UTC HRs of the UTI cohort were significantly higher in the subgroups categorized by comorbidity than in the non‐UTI cohort. Among these binary groups, the HR was higher in patients with hypertension (HR = 4.57, 95% CI = 3.04–6.87) and patients with hyperlipidemia (HR = 5.04, 95% CI = 2.62–9.69), but lower in patients with diabetes (HR = 3.20, 95% CI = 1.70–6.00), compared to their counterparts. However, adjusted HR was significantly higher only for patients with hypertension (adjusted HR = 1.37, 95% CI = 1.04–1.80).

When stratifying by time since UTI diagnosis, the incidence rates of UTC were higher in UTI patients than those in the non‐UTI cohort within 4 years since UTI diagnosis, with adjusted HR = 7.73 (95% CI = 4.54–13.2; P < 0.0001). The significant differences were also observed in 4–8 years and over 8 years follow‐up, although the HRs decreased over time (Fig. 1). Table 3 shows the incidence and crude and adjusted HR of UTC, renal pelvis and ureter, and bladder cancer associated with different sites of UTI. Lower UTI patients were generally at a higher risk of UTC than the non‐UTI comparison cohort (adjusted HR = 4.96, 95% CI = 3.74–6.57). Upper UTI patients had a relatively higher risk of renal pelvis and ureter cancer (adjusted HR = 4.26, 95% CI = 2.51–7.22). Conversely, lower UTI patients had a relatively higher risk of bladder cancer (adjusted HR = 5.68, 95% CI = 3.91–8.25).

Figure 1.

Probability of patients free of urinary tract cancer with urinary tract infection (UTI) (dashed line) or without UTI (solid line).

Table 3.

Incidence, crude, and adjusted hazard ratios of urinary tract cancer, renal pelvis and ureter, and bladder cancer between different sites of urinary tract infection (UTI) locations

Variables (ICD‐9‐CM code)	Events	PY	Rate †	Crude HR‡ (95% CI)	Adjusted HR § (95% CI)
Urinary tract cancer (188 and 189)
Non‐UTI	75	1 248 655	0.60	1.00 (Reference)	1.00 (Reference)
Upper UTI (590)	47	173 361	2.71	4.49 (3.11, 6.46)***	3.79 (2.62, 5.47)***
Lower UTI (595)	140	473 600	2.96	4.92 (3.71, 6.51)***	4.96 (3.74, 6.57)***
Renal pelvis and ureter cancer (189.1 and 189.2)
Non‐UTI	34	1 248 655	0.27	1.00 (Reference)	1.00 (Reference)
Upper UTI (590)	24	173 361	1.38	5.04 (2.99, 8.50)***	4.26 (2.51, 7.22)***
Lower UTI (595)	53	473 600	1.12	4.11 (2.67, 6.32)***	4.11 (2.66, 6.34)***
Bladder cancer (188)
Non‐UTI	41	1 248 655	0.33	1.00 (Reference)	1.00 (Reference)
Upper UTI (590)	23	173 361	1.33	4.03 (2.42, 6.71)***	3.41 (2.04, 5.72)***
Lower UTI (595)	87	473 600	1.84	5.59 (3.86, 8.11)***	5.68 (3.91, 8.25)***

*P < 0.05, **P < 0.01, ***P < 0.0001. †Incidence rate, per 10 000 person‐years; ‡related hazard ratio; §multivariable analysis including age, sex, and comorbidities. CI, confidence interval; HR, related hazard ratio; ICD‐9‐CM, International Classification of Diseases, 9th revision, Clinical Modification; PY, person‐years.

Discussion

The results from this population‐based cohort study showed a significant association between UTI and UTC risk. Further analyses indicated the presence of a correlation between sites of UTI and UTC. Furthermore, the increased UTC risks were consistently significant over time.

Data from the US National Cancer Institute's Surveillance, Epidemiology and End Results showed that overall cancer incidence rates of all racial and ethnic groups combined in the USA decreased by 0.8% per year between 2003 and 2007.19 However, this trend is reversed in Taiwan. Malignant neoplasm has been the leading cause of death for the Taiwanese general population since 1982, and the age‐adjusted incidence rate has steadily increased to 276 new cases per year and 100 000 diagnoses in 2008.20 This issue remains a challenge for Taiwan's public health, gaining the attention of the government; thus, more population‐based investigations regarding cancer‐preventive epidemiology continue to occur. The NHI program provides adequate data for these studies because of its comprehensive health coverage. The NHIRD contains hospital service records, ambulatory service records, and prescription claims data, allowing investigators to select specific study groups and matched comparison groups with confirmed representation of the underlying population groups. Our research group recently carried out a cohort study using the NHIRD to evaluate the relationship between end‐stage renal disease and subsequent cancer risk, and yielded positive results.21 This study used a similar design to determine whether UTI is associated with UTC risk.

According to 2008 cancer statistics, bladder cancer is the 8th and 13th most common malignancy among Taiwanese men and women, respectively.20 Numerous etiological factors are associated with the development of bladder cancer.22 Cigarette smoking is the most important risk factor in industrialized countries. Furthermore, specific bladder cancer‐causing chemicals and numerous occupational exposures to less‐defined specific agents have been identified. Arsenic has been identified as a bladder carcinogen in Argentina, Chile, and Taiwan.22, 23

Infections from members of the genus Schistosoma are responsible for a high incidence of bladder cancer in developing countries, especially those in the Middle East and in parts of Africa.10 Furthermore, non‐specific UTI was also suggested as a potential risk factor for the development of bladder cancer.8, 9 Most case–control studies support this and show a significantly higher risk for bladder cancer development in UTI patients,24, 25 but several studies revealed no significant difference between case and control groups.26, 27 Jiang et al.12 showed that a significantly reduced risk of bladder cancer existed among women with a history of bladder infection. Cytotoxicity from commonly used antibiotics for treating bladder infections is a proposed explanation. The results from this study are consistent with most case–control studies and suggest that patients with UTI have a higher risk of developing UTC. When the site of infection was divided into upper and lower urinary tracts, both groups had a significantly higher risk of UTC, compared to the control group, but trends were more compatible within locations. This indicates that upper UTI is associated with a higher risk of renal pelvis and ureter cancers, and lower UTI is associated with a higher risk of bladder cancer.

Jhamb et al.11 indicated that risk of bladder cancer was attenuated for infections diagnosed >1 year from the time of bladder cancer diagnosis. Gonzalez et al.28 observed a similar phenomenon where the increased risk existed only for infections starting 4 years or less before diagnosis and no statistically significant increase in risk for infections starting five or more years before. Furthermore, they suggested that the association of UTI with bladder cancer may not exist. However, this study indicated that the UTC‐free proportion of patients with and without UTI differed significantly even after 8 years of follow‐up, although the adjusted HRs decreased over time (Fig. 1).

Our study is substantiated by the population‐based design and its large and nationally representative sample. The results provide a realistic perspective for UTC incidence among UTI patients. However, there were certain limitations. First, information on potential confounding factors such as cigarette smoking and occupational exposure were unavailable for adjustment. Second, the NHIRD does not provide information regarding cancer histology type, and a definite relationship between bladder infection and bladder cancer has only been established for squamous cell carcinoma (as evidenced in cases of schistosomiasis)10 and spinal cord injury patients who are prone to chronic UTI.29 Kantor et al.30 showed a positive association between UTI and bladder cancer, particularly for squamous cell carcinoma. An earlier study from Taiwan found 6.6% of bladder cancer cases involved squamous cell carcinoma.31 This study was unable to ascertain whether more‐than‐average squamous cell carcinoma existed among bladder cancer patients. Despite a meticulous study design with adequate control of confounding factors, a crucial limitation is that bias may remain if there are unmeasured or unknown confounders. The UTI and UTC diagnosis data are otherwise highly reliable. These results are in agreement with those of most prior studies and suggest that patients with UTI are at a higher risk of developing UTC and the sites of infection and cancer are potentially predisposed. Increased UTC risks in UTI patients were observed to be consistently significant over time. Further large‐scale properly controlled prospective studies exploring possible underlying mechanisms are necessary.

Disclosure Statement

The authors have no conflicts of interest.

(Cancer Sci 2013; 104: 619–623)