Risk of extrahepatic cancer in a nationwide cohort of hepatitis C virus infected persons treated with direct‐acting antivirals

Direct‐acting antivirals (DAAs) against HCV have an immune modulatory effect, this could possibly lead to a decreased tumour control. We, therefore, aimed to assess the risk of extrahepatic cancer (EHC) during and the first years after DAA treatment.


| INTRODUC TI ON
Chronic HCV infection is a global health problem with an estimated 71 million people infected. 1 After 20 years with chronic HCV infection, 20%-25% develop liver cirrhosis, with an increased risk of HCC, liverfailure and liver-related death. 2 Chronic HCV infection is also associated with an increased risk of some extrahepatic cancers (EHC) such as non-Hodgkin lymphoma and intrahepatic cholangiocarcinoma. [3][4][5][6][7][8] The treatment of chronic HCV infection was revolutionised in 2014 with the introduction of direct-acting antivirals (DAAs) with cure rates of ≥95% and mild or no side effects. 9 Previous interferon (IFN)based therapies had low cure rates and serious side effects, and were not suitable for patients with conditions such as hepatic decompensation or some psychiatric disorders. 10 With new treatment options, the World Health Organization (WHO) has set up a goal to eliminate hepatitis C by 2030 and many countries have scaled up DAA treatment, with approximately 5 million treated worldwide by the end of 2017. 11,12 Within a couple of years after the introduction of DAAs, there were surprising reports of both occurrence and recurrence of HCC in patients recently treated with DAAs. 13,14 A substantial proportion of the detected HCCs had aggressive patterns and only palliative treatment could be offered. In contrast, other studies have shown long-term clinical benefits of DAA treatment, including decreased risk for HCC, lower risk of decompensation and a reduction in liverrelated mortality. [15][16][17][18] As a result, there have been different opinions about DAA treatment in relation to the risk for HCC and it has been proposed that further studies of safety are required. [19][20][21] The immunological changes caused by DAAs and a possibly decreased tumour control have raised a question about the risk for EHC in relation to DAA treatment. 21 A recent study by Kim et al, demonstrated an increased mortality from EHC in the DAA era in a large population cohort, but HCV treatment data were lacking in that study. 22 Other publications have described an unexpectedly high occurrence or recurrence of EHC in small cohorts of DAAtreated patients. 23,24 Large studies of EHC risk in relation to DAA treatment per se are still missing. In Sweden, the personal identification number (PIN) of all residents and the national registers with high coverage and prospectively collected data present a unique opportunity to study this.
The aim of this nationwide population-based register study was to study the risk of EHC during and shortly after DAA treatment by comparing the risks for EHC in DAA treated with IFN-treated and -untreated HCV-infected patients. Additionally, we compared these groups with matched controls without HCV from the general population.

| Study population and data sources
All Swedish residents have a personal identification number (PIN) which is used by all authorities, health care and in national registers, records with all HCV notifications from 1990 to 2015, including information on date of notification (assumed to be the date of HCV diagnosis) and route of transmission, was extracted from the Public Health Agency, as described in previous studies. [26][27][28] This dataset was sent to Statistics Sweden (SCB), where duplicates and notifications with incomplete PIN were excluded (n = 5174) and information on date of death, emigration/immigration and country of origin were extracted ( Figure 1). 29 Individuals who did not live permanently in Sweden at the time of HCV-notification (n = 1935), and individuals co-infected with HBV-infection (n = 4038) were excluded. The remaining HCV-cohort comprised 53 002 individuals. The reported routes of transmission were unknown/missing in 50%, injection drug use in 40%, transfusion of blood/blood products before 1992 in 5%. 30 From 1992 injection drug use is the assumed route of transmission in the majority of new infections. 25 An individually matched comparison-cohort without HCV diagnosis from the general population was obtained from SCB. For each HCV-infected person, 10 individuals, alive on the date of notification and matched for sex, birth year and county of residence, were collected. All comparators with inconsistent data, no observation-time, past/present EHC, liver cancer, or liver transplantation at inclusion, or who had died prior to or at the same day as inclusion date were excluded.

| Linkage with national registers
The data from the HCV-infected cohort and the matched comparisoncohort were linked to national registers with prospectively collected data held at the National Board of Health and Welfare; the Prescription Register (PrR), the Patient Register (PR), the Cancer Register (CR) and the Cause of Death Register (DR). 31  For sensitivity analyses, EHC diagnoses from the PR and DR were identified (ICD-10) and added when missing in the CR. The accuracy of the PR and DR is lower than the accuracy of the CR (with high proportion of histological diagnoses), therefore EHC diagnoses from the PR were included only when registered at least twice. 34,35

| Treatment groups
The three therapy groups, DAA-treated, IFN-treated and the untreated group were identified among eligible subjects in the total HCV-cohort (Flow chart, Figure 1). Each individual could only be included in one group. Everyone with a liver cancer, EHC, or liver transplantation before the start of follow-up was excluded.
The DAA-treated group consisted of patients treated with IFNfree regimes with 2nd/3rd wave DAAs. Those who had been treated

| Charlson Comorbidity Index
Charlson Comorbidity Index (CCI) is a method of categorising comorbidity based on ICD-codes in outpatient and inpatient discharge diagnoses, with each comorbidity category having an associated weight, based on the adjusted risk of mortality or resource use. The sum of all weights results in a single comorbidity score for the patient. 36 Data for calculating CCI for HCV-infected individuals and comparators were retrieved from the PR. The CCI-diagnoses were included from 5 years before start of the study, until end of followup. Malignancies were excluded from the CCI-diagnoses in this study.

| Ethics
The study was approved by the Regional Ethical Review Board in Stockholm, Sweden. The linkage of the register-files was handled by the Public Health Agency, Statistics Sweden and the National Board of Health and Welfare. Only anonymised data were used for analyses. and for the matched general population, and log-rank tests to compare EHC-free follow-up time between the groups.

| Statistical analysis
The data management and statistical analyses were carried out using the statistical software SAS® (Version 9.4, SAS Institute Inc). A two-sided P < 0.05 was considered statistically significant.

| Comparison of risk in DAA treated vs other treatment groups
The results of the Cox regression analyses by treatment group, for the whole group and by sex, are presented in Kaplan-Meier curves of EHC free survival time for the three treatment groups and the DAA-matched controls, divided in three age groups, are demonstrated in Figure S1A-C. Log-rank tests showed statistically significant risk difference for the DAA group vs untreated group and for the untreated vs IFN group in the age-group 50-64 (P = 0.003 and <0.001 respectively), but not in the other age groups.   United States and found an 11% decreased incidence when comparing patients treated with any kind of HCV therapy with untreated patients. 37 The association was not observed with only DAA treatment due to the short follow-up time for this group (12.7 months).

| D ISCUSS I ON
Treatment response was not known in this study. Another study showed that IFN-induced SVR was significantly associated with a reduction of the risk of lymphoma, multiple myeloma, MGUS and haematological malignancies combined, but these associations were not observed with DAA-induced SVR during a mean follow-up of

TA B L E 4 Cox regression models
showing HR for extrahepatic cancer in hepatitis C patients treated with directacting antivirals (DAA) or interferon (IFN), compared to matched control groups from the general population 2.9 years. 38 Neither IFN-induced SVR nor DAA-induced SVR was associated with the risk of colon cancer or prostate cancer. A strong association between viral eradication and a better outcome of HCVpositive B-cell non-Hodgkin lymphoma has also been shown in a meta-analysis, but due to lack of data, separate analyses on DAAtreated patients were not possible. 39 The three treatment groups in our study differed in several aspects. The DAA group had a 10-year higher mean age, higher CCI and more liver cancers than the other groups. This could be ex-  Our study showed an about 1.5-fold higher risk of EHC for both DAA-and IFN-treated men compared with matched comparators without HCV diagnosis from the general population (statistically significant for DAA treated). This could possibly be related to a lifestyle with higher alcohol and tobacco consumption than the matched comparison cohort, such information was not available since these risk factors seldom are registered as ICD-codes in clinical praxis. Additionally, surveillance bias of the DAA-and IFN-treated groups compared with the uninfected comparison group could possibly contribute to this finding. No decline of the increased risk for EHC in DAA treated compared with the general population was seen after 3 years of follow-up, even though ≥95% of DAA treated in Sweden have been cured. 40 This implies that viral eradication does not diminish the elevated risk of EHC during the first years after treatment. A French prospective study found that HCV cirrhosis was associated with a higher risk of EHC compared to the French general population, even after cure with

IFN. 41
The reports about a possible increased risk of early recurrence and more aggressive HCCs after DAA treatment raised a concern about the safety of DAAs. 13,14 Many studies of DAAs and HCCs have been conducted the last years, and most of them showed no elevated HCC-risk after DAA treatment, but a diminishing HCC-risk over time. 16,17 Hepatitis C virus infects not only hepatocytes but also extrahepatic cells and could induce alterations of the immune system and local tissue microenvironment, with increased risk of various extrahepatic neoplasia. 7 Immunological studies have shown that chronic HCV patients have an altered natural killer (NK) cell phenotype, which is exacerbated during IFN therapy, but normalised with DAA therapy. The fast suppression of the virus reduces the level of inflammatory cytokines in DAA treated and is associated with a suppression of NK cells, which may reduce immunosurveillance of precancerous clones. 42,43 It has also been shown that treatmentinduced HCV clearance decreases the strength of HCC-specific CD8 T-cell response in cirrhotic patients and may fail to recover after DAA therapy. 44 Interferons have a well-known role in promoting antitumor responses and have been extensively used for the treatment of several cancers, and there are theories of a protective effect of IFN on HCC development that is not seen with DAA treatment. 45 Strengths in our study are the nationwide approach and the rather large sample size. Other advantages are that the data were prospectively recorded, and the high completeness of the Swedish HCV surveillance register with notifications from both the clinician and the diagnosing laboratory, as well as the other national registers in Sweden. The use of PINs provides a unique opportunity to conduct register studies in Sweden, since many different registers can be combined and provide extensive information on each individual

included.
A possible limitation is the lack of data regarding cure after treatment. This is a concern for the IFN group with cure rates around 50%, varying by HCV genotype and stage of liver fibrosis, in contrast to ≥95% in the DAA group. Cure of HCV results in a lower incidence of HCC and liver-related mortality, and longer survival may result in more EHC in the long term due to ageing without liver-related complications. However, the cure rate may not be an important factor if cancer is suspected to arise from immune-modulation early after initiation of therapy, which has been the concern. Another limitation is the lack of information about IFN treatment prior to 2005, before the start of the PrR. The impact of this limitation is probably low, since inclusion in the study started in 2008 and all individuals with a malignancy prior to inclusion were excluded. Additionally, the lack of data on smoking and alcohol consumption is a limitation when studying the risk of EHC in HCV-infected individuals compared with the general population. Nevertheless, in this study, the main objective was to investigate if there was an increased risk of EHC associated with DAA treatment by comparing the HCV treatment groups, assuming similar alcohol and tobacco use. The comparison with the general population for both DAA and IFN treated was added to illustrate the risk from another point of view, but still with the focus on DAA treatment in relation to IFN treatment.
This study estimates the EHC risk early after DAA treatment with follow-up time limited to a maximum of 3 years. This time frame is though sufficient to study the early incidence of cancer, which is important since previous reports suggested an increased risk for HCC during or soon after treatment start. In our study, sensitivity analyses with follow-up limited to 1 and 2 years were performed, but no statistically significant difference was found at any time-point.
However, studies with longer follow-up, to estimate the long-term risk, are needed when DAAs have been used for a longer time.
In this nationwide register study of people with HCV diagnosis, there was no statistically significant increased risk for EHC after DAA treatment when adjusted for age and comorbidity. This suggests that DAA treatment is safe in respect to EHC early after treatment, but confirmatory studies and studies with longer follow-up time are needed to fully explore this issue. The study indicated an increased risk of EHC in people with an HCV diagnosis compared with members of the general population without HCV. This association remained after cure from HCV and further studies of the ageing population with a history of HCV infection would be of interest.

CL was supported by research ALF grants from Region Örebro
County (OLL-841771 and OLL-930212). AD was supported by research ALF grants from Region Örebro County (OLL-683801). The study was supported by Nyckelfonden, Region Örebro County, Sweden (OLL-691511), Region Stockholm County and Cancerfonden without involvement in any part of the study, manuscript or decision to submit.

CO N FLI C T O F I NTE R E S T
A-S. D. has served as a speaker and/or consultant for AbbVie, BMS, Gilead and MSD. SA has served as a speaker and/or consultant for AbbVie, BMS, Gilead, Janssen, MSD/Merck, Medivir, Roche and Tillotts Pharma and has received research grants from AbbVie and Gilead. CL has served as a speaker for MSD. SM and DB reports no conflicts of interest.

AUTH O R S ' CO NTR I B UTI O N S
A-SD, SA and CL contributed to the concept of the study. All authors contributed to the design of the study and interpretation of data. A-SD, SA and CL made the data collection. DB performed the statistical analyses. The manuscript was drafted by CL, and critically revised and approved by all authors.

PE E R R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1002/ygh2.456.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.