Impact of Hepatitis C Virus Treatment on the Risk of Non-hepatic Cancers Among Hepatitis C Virus-infected Patients in the US

Wei Wang; Vincent Lo Re III; Yi Guo; Hong Xiao; Joshua Brown; Haesuk Park


Aliment Pharmacol Ther. 2020;52(10):1592-1602. 

In This Article

Materials and Methods

We conducted a retrospective cohort study using the IBM MarketScan Commercial Claims and Encounters and Medicare Supplemental and Coordination of Benefits Database (IBM® MarketScan® Research Databases) from January 2005 to December 2016. This claims database contains de-identified person-level information of diagnoses, procedures, and prescriptions for over 180 million individuals from 1 January 2005 to 31 December 2016. Approximately 10% of these individuals have Medicare Supplemental insurance coverage. The distributions of patients' sex, age and geographic location in these data are representative of Americans covered by commercial health insurance. MarketScan data capture the continuum of patients' care and medical expenditures in multiple settings, including physician office visits, hospital stays and pharmacies, thus enabling longitudinal studies.[23]

Study Population

Patients with newly diagnosed chronic HCV infection were identified using the International Classification of Diseases, Clinical Modification codes (ICD-9-CM codes: 070.44, 070.54, 070.70, 070.71 and V02.62; and ICD-10-CM codes: B18.2, Z22.52, B19.20 and B19.21) (Table S1). We considered a patient to be chronically infected with HCV if they had one in-patient claim or two out-patient claims of HCV infection on separate days within 1 year. These diagnoses have been shown to be highly, with a positive predictive value of 88% (95% confidence interval (CI): 82.5%-92.2%) for identifying HCV infection.[24] We considered the initial HCV diagnosis date as the index date. We included patients who met the following criteria: (a) aged 18 years or older at the time of the first HCV diagnosis, (b) continuously enrolled in the health plan for 12 or more months before the index date (to ascertain baseline covariates and previous medication use), and (c) continuously enrolled in the health plan for at least 12 months after the index date (to allow sufficient time to evaluate the non-hepatic cancers of interest). Follow-up for all patients began on the index date. The 12 months prior to the index date represented the baseline period. We excluded patients who had a history of HCV treatment or a history of any primary or metastatic cancer (ICD-9-CM: 140.xx-239.xx, ICD-10-CM: C00-D48) during the baseline period.[25]


The primary study outcome was an incident non-hepatic cancer, including cancer of the lung, colon/rectum, non-Hodgkin's lymphoma, head/neck, oesophagus, pancreas, prostate or breast. We defined non-hepatic cancer as a composite outcome that was determined by ICD-9/10 diagnosis codes (ICD-9-CM: 140.xx-195.xx, except 155.x primary liver cancer; and ICD-10-CM: C00-D48, except C22 primary liver cancer; see Table S2 for list of diagnoses). We considered patients to have developed an incident cancer if they had at least two diagnoses of a specific cancer type in a 2-month period in in-patient and out-patient profiles.[25] We considered a patient's first non-hepatic cancer diagnosis date after their index date as the date of cancer onset. As secondary outcomes, we evaluated the non-hepatic cancers separately.

HCV Treatment and Time-dependent Exposure

HCV Treatment Initiation (TX Initiation). Patients were considered to have received HCV treatment if they initiated one of the following HCV therapies: (a) dual therapy, a combination therapy of interferon and ribavirin (ie interferon alpha, interferon beta, peg-interferon alpha-2a or peg-interferon alpha-2b with or without ribavirin); (b) triple therapy, a combination therapy of peg-interferon and ribavirin plus first-generation direct-acting anti-viral agents (DAAs) (eg boceprevir or telaprevir) or second-generation DAAs (eg sofosbuvir or simeprevir); or (c) all-oral therapy, including sofosbuvir, ledipasvir/sofosbuvir, sofosbuvir plus simeprevir, ombitasvir/paritaprevir/ritonavir, sofosbuvir plus daclatasvir, ombitasvir/paritaprevir/ritonavir plus dasabuvir, elbasvir/grazoprevir, and sofosbuvir/velpatasvir with or without ribavirin. The date of HCV treatment initiation (TX initiation) was considered as T1 (Figure 1). We defined T1 as the date of the first interferon prescription for patients who initiated dual therapy or the date of the first DAA prescription for patients who initiated triple or all-oral therapy. HCV patients who had not received any HCV therapies were considered as untreated (No Tx).

Figure 1.

Study design. Abbreviations: HCV, Hepatitis C virus; TX, Treatment

Minimally Effective Treatment (Effective TX). We considered patients who initiated HCV treatment to have completed a minimally effective treatment (Effective TX) if they received one of the following HCV treatment regimens: 16 weeks of dual therapy,[15] 12 weeks of triple therapy with a first-generation DAA,[16] 8 weeks of triple therapy with a second-generation DAA,[17] or 8 weeks of all-oral therapy.[18] The date that patients completed a minimally effective treatment was marked as T2 (Figure 1). We allowed a 30-day grace period to determine patients' treatment discontinuation. If patients had early refills, the cumulative days of supply were used to calculate T2.

Time-dependent Exposure. Based on patients' initiation and completeness of minimally effective treatment, a time-dependent exposure method was employed in the analysis. Patients were classified as having a time-dependent treatment status of "No TX", "TX initiation", or "Effective TX" during the follow-up time. Patients' follow-up time started on the index date (HCV diagnosis) and continued until: (a) first occurrence of a non-hepatic cancer, (b) switching of HCV treatment regimen, (c) end of health plan enrollment, or (d) 31 December 2016 (the end of study data), whichever came first.


We identified patients' age, sex, commercial insurance type, Charlson Comorbidity Index (CCI) score, relevant medical comorbidities and medications that increase or decrease the risk of some cancers. Patients' medical comorbidities were defined by one in-patient or two out-patient ICD-9-CM and/or ICD-10-CM diagnoses (recorded on separate dates within 1 year) of alcohol dependence/abuse, smoking-related diagnosis, drug use disorder, cirrhosis, decompensated cirrhosis, chronic kidney disease, diabetes, HIV infection, chronic hepatitis B virus infection, obesity, rheumatoid arthritis, chronic pancreatitis, human papillomavirus infection and inflammatory bowel disease (see Table S3 for lists of diagnoses for each condition). We also identified diagnoses that were considered contraindications to peg-interferon-based therapy, specifically schizophrenia, depression, seizures, pregnancy, non-liver organ transplant, anaemia and retinopathy.[20,26–36] Medications with preventative or hazardous effects on cancer were identified using the National Drug Codes (NDC) and the unique therapy class in the database (Table S4), including statin, proton pump inhibitor (PPI) and hormone therapies.[10,27] Among all HCV-infected patients, to adjust the potential impact of different healthcare utilisation behaviour on the cancer outcome identification, we also identified the mean number of in-patient visits and out-patient office visits in the 12 months prior to the index date.

Statistical Analysis

Baseline characteristics were compared between patients with no treatment and patients who received dual therapy, triple therapy or all-oral therapy by using Student's t test for continuous variables and chi-square tests for categorical variables. The number of non-hepatic cancer events and the person-year of observation were determined for each treatment status group and were used to calculate the incidence rates of non-hepatic cancers (number of events/100 000 person-years). Multivariable, time-varying Cox proportional-hazards regression models were used to examine the impact of treatment initiation and minimally effective treatment on the risk of developing non-hepatic cancers by incorporating time-dependent exposure methods. Patients' sex, commercial insurance type, CCI, and health service utilisation at baseline were adjusted as time-fixed covariates, while age, HCV disease duration, clinical conditions (eg alcohol use disorder, drug use disorder, cirrhosis, decompensated cirrhosis, HBV, HIV, diabetes, obesity, smoking, IBD, rheumatoid arthritis, HPV, chronic pancreatitis), and medication use (eg statin use, PPI, hormone therapy) were adjusted as time-dependent covariates in the models. The covariates adjusted in the primary composite non-hepatic cancer outcome model and each secondary specific cancer outcome model are listed in Table S5. For all models, hazard ratios (HRs) and their 95% CIs of non-hepatic cancers were measured.

We conducted four subgroup analyses. First, we conducted subgroup analyses stratified by sex, age categories (<50 years; 50–64 years; ≥65 years), cirrhosis and diabetes mellitus, each of which may modify the risk of non-hepatic cancer among HCV-infected patients.[10,20,26] Second, we stratified patients by HCV therapy type (ie dual therapy, triple therapy or all-oral therapy) to examine the therapy-specific effects on non-hepatic cancers. Third, in the analysis for the impact of triple therapy treatments, we further stratified patients by first-generation triple therapy or second-generation triple therapy. Finally, we conducted a sensitivity analysis in the databases of 2011–2016 and 2013–2016, identifying patients from the time period after the approval of triple or all-oral therapies, respectively, to ascertain comparable follow-up times for treated HCV patients and untreated HCV patients.

The proportionality of hazards was evaluated prior to fitting all models. All analyses were performed using SAS 9.4 (SAS Institute Inc, Cary, NC). P value of 0.05 was used to evaluate statistical significance.