Trends in Hepatocellular Carcinoma Incidence and Risk Among Persons With HIV in the US and Canada, 1996-2015

Key Points Question Has the incidence rate or risk for liver cancer among people with HIV increased with the widespread availability of combination antiretroviral therapy? Findings In this cohort study of 109 283 people with HIV in North America, the incidence rate of liver cancer increased between the early and modern combination antiretroviral therapy eras. People with HIV coinfected with hepatitis B virus and/or hepatitis C virus, those with higher HIV RNA levels and lower CD4 cell counts, and persons who inject drugs had a higher liver cancer risk. Meaning These findings underscore the importance of achieving immune recovery along with monitoring for and long-term management of viral hepatitis among people with HIV.

Participants no longer contributed person-time at the first date of HCC diagnosis, death, December 31, 2015, or end of cohort-specific cancer ascertainment.

HCC Case Definition
HCC diagnoses (451 cases) were collected by extraction from medical records using a Web-based standardized abstraction protocol for 415 cases; 36 cases were identified by linkage to cancer registries. 4 Among non-cancer registry cohorts, all reported cases were verified by trained medical record abstracters under the supervision of physicians. Histopathology, laboratory, and imaging results were used to adjudicate all cases. The date of first confirmed HCC diagnosis was used for analysis.

Other Covariates
Sociodemographic information (age, sex, and race) and HIV transmission risk (men who have sex with men [MSM], people who inject drugs [PWID], heterosexual contact, and other or unknown) were collected through self-report or medical record review at enrollment. Individuals who were both MSM and PWID (1229 individuals [1.1%]) were categorized as PWID. Antiretroviral regimens were characterized at beginning of each cART era. CD4 cell counts (cells per microliter) were measured at the time closest to entry of each cART era or beginning of cancer observation within that period. HIV RNA levels (copies per milliliter) were measured 2 years (18-30 months) before HCC diagnosis or censoring (loss-to-follow-up or end of observation). FIB-4 (fibrosis-4) scores were calculated using standard methods based on studies conducted in patients with chronic hepatitis, with a value greater than 3.25 considered advanced liver fibrosis. 25

Statistical Analysis
The χ 2 or Wilcoxon test (2-sided) was used to compare differences in baseline characteristics by subsequent diagnosis of HCC. Significance was set at P < .05. Poisson regression models were used to estimate HCC IRs per 1000 person-years. In case of overdispersion in models, generalized estimating equation models were used as an alternative. We compared IR ratios (IRRs) with 95% CIs by viral hepatitis coinfection groups, transmission risk groups, CD4 cell counts (Յ500 vs >500 cells/μL), and HIV RNA levels (<500 vs Ն500 copies/mL). IRRs were adjusted for age group (<40, 40-49, 50-59, and Ն60 years), sex, race (White, Black, and other or unknown), and cohort; additional adjustment for viral hepatitis coinfection was performed when appropriate. Age-specific cumulative incidence of HCC by viral hepatitis coinfection group was calculated. Sensitivity analysis compared models with and without accounting for competing risks of death. Statistical analyses were conducted using SAS statistical software version 9.4 (SAS Institute) and completed in March 2020.

JAMA Network Open | Infectious Diseases
Trends in HCC Incidence and Risk Among Persons With HIV in the US and Canada, 1996-2015

IRs, IRRs, and Cumulative Incidence of HCC by HBV-HCV Coinfection
There was no substantial change in crude IRs of HCC across 3 eras among the HIV monoinfection group (eTable 3 in the Supplement). Rates of HCC among PWH coinfected with HBV were elevated in the early cART era and then increased in the modern cART era, whereas rates substantially increased across 3 eras among PWH coinfected with HCV (early cART: IR, 0. 34 Table 2 and eTable 3 in the Supplement). HBV or HCV coinfection was associated with a more than 20 times higher incidence, and triple infection was associated with nearly 40 times higher incidence of HCC compared with HIV monoinfection in the modern cART era. Coinfection with HBV and/or HCV was also associated with higher age-specific cumulative incidence of HCC compared with HIV monoinfection (Figure 1).
Results from sensitivity analysis in a subset of the cohort (35 189 participants [32.2%]) further adjusted for alcohol and smoking consumption were consistent with primary models (eTable 4 in the Supplement).
We compared age at HCC diagnosis with the underlying age distribution of each coinfection group (Figure 2

IRs of HCC by HIV Disease Severity
In the modern cART era ( Table 3), PWH with recent HIV RNA levels greater than or equal to 500 copies/mL had an 80% increased HCC IR (IRR, 1.8; 95% CI, 1.4-2.4) compared with those with HIV RNA levels less than 500 copies/mL. Similarly, PWH with lower CD4 cell counts (Յ500 cell/μL) had 30% increased HCC IR vs PWH with higher CD4 cell counts (IR 1.3; 95% CI, 1.0-1.6). Both outcomes

JAMA Network Open | Infectious Diseases
Trends in HCC Incidence and Risk Among Persons With HIV in the US and Canada, 1996-2015 were independent of viral hepatitis coinfection and covariates. Higher HCC rates associated with more advanced HIV disease severity were not observed in earlier calendar periods (eTables 6, eTable 7, and eTable 8 in the Supplement).

HCC Rates by HIV Transmission Group
PWID had higher HCC incidence and increased IRRs compared with MSM (eTable 9 in the Supplement). We hypothesized that this was largely associated with differences in viral hepatitis coinfection. eFigure 4 in the Supplement demonstrates the distribution of known HIV transmission risk groups by viral hepatitis status. PWID represented most of the HCV coinfection group, whereas MSM represented most of the HBV coinfection group.

Discussion
We examined HCC temporal trends and risk factors among PWH in North America over a 2-decade period, representing one of the largest and longest prospective studies with validated cancer outcomes of HCC in the context of HIV. HCC is among the most lethal cancers worldwide and is one of the few malignant entities with an increasing incidence 26 and mortality in the US in recent decades. 27 PWH had a  America. The increase in both IR and number of cases of HCC was most noticeable among individuals aged 50 years or older, which suggests that population age structure change and survival benefits are the major contributing factor for the increased IR. The use of cART has prolonged survival among PWH, which suggests the burden of HCC might further increase as the population continues to age with longer exposure to both HIV and chronic liver diseases. When comparing the modern vs the early cART eras, the IRRs of HCC remained significant even after controlling for age groups, suggesting that additional factors could be associated with the increased incidence. The increased rate might also be attributed to other nonviral causes (eg, alcohol abuse, obesity, diabetes, and nonalcoholic fatty liver diseases). In the general population, the prevalence of alcohol abuse and nonalcoholic fatty liver diseases has been increasing in the US concurrent with our study period. 31,32 These nonviral risk factors could act synergistically with viral hepatitis and might play a role in the increase.
PWH coinfected with HBV and/or HCV had substantially higher HCC rates and had earlier HCC diagnoses compared with HIV monoinfected persons, and this difference was most pronounced in the modern cART era. These findings are consistent with the previous observation that viral hepatitis coinfection is associated with a higher risk of end-stage liver disease. 33 The incidence of HCC among PWH coinfected with HCV was higher in our study than that observed in the study by Gjaerde et al 30 times higher among PWH coinfected with HCV and more than 100 times higher among PWH coinfected with HBV). 26 PWH with viral hepatitis coinfection have elevated HCC incidence rates that are approaching the level suggested to benefit from HCC surveillance in other selected populations. 35 Data from the current study warrant further investigation, including randomized clinical trials to evaluate the efficacy of more active cancer surveillance among PWH.
As many as 25% of PWH are coinfected with chronic HCV 36,37 and 8% to 15% of PWH are coinfected with chronic HBV 9 globally because of shared routes of transmission. 37 The substantial contributions of chronic HBV and HCV infection to prolonged liver inflammation, regeneration, fibrosis, and ultimately HCC have been well described elsewhere. 38 Several ART medications (eg, lamivudine, emtricitabine, tenofovir, and emtricitabine) are also active against HBV. Through effectively controlling both HBV and HIV viral replication, these therapies could improve liver-related outcomes. 8,46 However, despite the widespread use of HBV-active ART, we previously reported that a substantial number of PWH with HBV were not receiving appropriate regimens. 33 Given the substantial prevalence of HBV coinfection and increased risk of HCC, vaccination against HBV and/or testing for protective antibodies remains vital among PWH. Broader uptake and early initiation of HBV-active ART among PWH coinfected with HBV also remains a critical secondary prevention strategy. 47-49

JAMA Network Open | Infectious Diseases
Trends in HCC Incidence and Risk Among Persons With HIV in the US andCanada, 1996-2015 Previous studies 50,51 suggested that a longer duration of HIV viremia and lower CD4 cell counts (<200 cells/μL) were associated with higher risk of HCC. Our findings suggest that even recent exposure to higher viral load and to less pronounced immunosuppression (CD4 cell counts Յ500 cell/μL) could also be associated with higher HCC risk. PWH suffer from persistent immune dysregulation and chronic inflammation, 14

Limitations
This study has limitations that should be addressed. First, although NA-ACCORD is representative of patients with HIV successfully linked to health care in North America, we were not able to make inferences about PWH who are not receiving health care. Second, although our observation time and sample size are large for an HCC study, granular data on potential risk factors (eg, natural clearance vs treated vs chronic HCV, alcohol, smoking, or obesity) were not available across all cohorts, making investigation of other important etiologic factors a limitation. However, results from a sensitivity analysis within a subset (32.2% of the cohort) with data on alcohol and smoking were consistent with our primary observations. Third, not all participants had been tested for HBV and HCV in NA-ACCORD, and testing for HBV and/or HCV in clinical cohorts was based on clinical decision. We might have misclassified some HCV-or HBV-coinfected patients in the HIV monoinfection group and underestimated the risk of HCC in the HBV and/or HCV coinfection groups. Fourth, completeness of HCC case ascertainment has been estimated to be robust at individual sites but not uniform across all sites, where sites could use either the cancer registry or electronic medical record ascertainment or both methods for case definition. However, the NA-ACCORD Data Management Core has conducted quality assessment and control of case definitions to minimize any differences through the harmonization process. 21,22 Fifth, DAAs uptake for hepatitis C treatment has been limited in the current database and rate of interferon-based regimens were low in NA-ACCORD before the DAA era. Therefore, we were not able to draw further implications for HCC risk among PWH receiving DAAs treatment. With further data accumulation in NA-ACCORD, this will be a focus of future analyses.

Conclusions
In conclusion, we used a large HIV collaborative study design to determine rates for HCC among PWH in North America in the 20 years since the availability of cART. The increases in HCC rates are of concern and may lead to continuing increases in HCC-related mortality in coming years. PWH coinfected with viral hepatitis with higher HIV RNA levels or lower CD4 cell counts or who contracted HIV through injection drug use had higher HCC rates. These findings underscore the importance of achieving immune recovery and monitoring for and long-term management of viral hepatitis among PWH.