Study design

We conducted a nationwide follow-up study to examine the spectrum of malignancies among Chinese adult PLHIV. To do this, we used routinely collected data for PLHIV who had been diagnosed with HIV infection on or before 31 December 2011 and were still alive and in care as of 1 January 2008. All participants contributed observed time from 1 January 2008 (i.e., the study start date) to 30 June 2012 (i.e., the study end date) and cancer cases were identified during this follow-up time.

Setting and data source

The data used for this study were collected during the normal operation of China’s HIV/AIDS Surveillance System and NFATP. All data related to China’s HIV response efforts are collected and maintained in a real-time, web-based repository called the HIV/AIDS Comprehensive Response Information Management System (CRIMS), which has been described elsewhere [31]. In brief, CRIMS contains records for all individuals diagnosed with HIV in China, and these records include contact and demographic information, testing and baseline clinical information, dates and details of all follow-up visits, and ART regimens and related assessments, as well as other health-related information (co-infections, co-morbidities) and dates and causes of death. After a diagnosis of HIV infection, but prior to the initiation of ART, PLWH are followed up once every 6 months. After ART initiation, patients are followed up 4 times in the first 3 months, and once every 3 months thereafter.

Eligibility criteria

Study inclusion criteria were: 1) being ≥15 years of age, and 2) having been diagnosed with HIV on or before 31 December 2011. All patient records in CRIMS meeting these study inclusion criteria were extracted and then screened against exclusion criteria. Study exclusion criteria were: 1) having died or been lost to follow-up prior to the start of the study on 1 January 2008, or 2) not having attended at least one follow-up visit prior to 30 June 2012, the date all data were extracted. All remaining patients were included in the analysis.

Case identification

Cancer cases were identified using reports of AIDS-related complications or causes of death at each follow-up visit. We categorized cancer cases based on the International Classification of Disease for Oncology, 3rd Edition (ICD-O-3) [32]. However, because approximately half of lymphoma records were not specified as to their type, we combined all cases of NHL, Hodgkin lymphoma (HL), and all other lymphomas together into a single category. This group was treated as a single cancer type in all analyses.

Observed time

The start of observed time was defined as either the study start date, 1 January 2008 (for those already diagnosed with HIV), or the date of HIV diagnosis, whichever was later. The inferred date of cancer incidence was defined as the median date between the date of cancer diagnosis and the date of the previous follow-up visit. Participants with inferred date of cancer incidence after the end of the study, 31 December 2011, were treated as cancer-free during the study period. For cancer-free individuals, the end of follow-up was defined as either 1) the date of last follow-up for those who were lost to follow-up, 2) the date of death, or 3) the end of the study for those who were still living. Observed time was calculated as the difference between first and last observation, expressed in person-years (PY).

Statistical analysis

Characteristics of participants were presented as number and percent. Sex-specific cancer incidences were calculated for each malignancy site/type, and age-standardized incidence rates (ASIR) and standardized incidence ratios (SIR), along with their 95% confidence intervals (CIs) were calculated. Direct standardization was used to calculate ASIR. Two different standard populations were used: ASIR-China (ASIRC) was calculated using China’s 2010 census data [33], and ASIR-world (ASIRW) was calculated using the World Health Organization (WHO) standard world population for 2002 to 2025 [34]. ASIRC and ASIRW were expressed per 100,000 PY. SIR was obtained by dividing the numbers of observed cases (numerator) by the expected numbers of cases (denominator). Expected cancer frequencies were determined using the observed sex- and 5-year, age-specific incidence rates from China’s National Cancer Registry in 2008 [35]. All CIs presented were calculated assuming a Poisson distribution. SAS software version 9.3 (SAS Institute Inc., USA) was used for all analyses.

Ethical considerations

This study was approved by the Institutional Review Board of the National Center for AIDS/STD Control and Prevention, China CDC, and by the Institutional Review Board of the University of California, Los Angeles. No informed consent was sought since all individuals diagnosed with HIV in China sign informed consent at the time of their initial entry into CRIMS, which includes the future use of their data for epidemiological study. All records, once extracted, were de-identified to ensure confidentiality.

HHV-8 Infection

KS is well-known to be caused by HHV-8, and PLHIV are commonly infected with HHV-8. In the pre-ART era, KS incidence was observed to be up to 2,000-fold greater among PLHIV, and the more severe the immunodeficiency, the higher the likelihood of KS occurrence [8]. In our study, 171 cases of KS were observed, which represented only 4.5% of all cancer cases observed. Not surprisingly, KS SIR was high—2,639.8 among males and 1,593.5 among females—as KS in the absence of HIV is very rare in Mainland China. However, the ASIRC of KS (23.9 per 100,000 for males and 14.2 per 100,000 for females) was lower than in the Taiwan study [29], and in many previous studies in the United States, Europe and sub-Saharan Africa [5–7,10–26,37]. The low prevalence of HHV-8 infection in Mainland China may be one cause of lower KS incidence rate [38,39]. Previous studies have shown that the incidence rate and SIR of KS has dropped dramatically since the introduction of ART [5,12,18,21], and we expect a similar trend to develop in China, since ART coverage is still expanding under the NFATP as China strives to meet the Joint United Nations Programme on HIV/AIDS (UNAIDS) 90-90-90 Targets [40,41].

EBV Infection

EBV infection is known to be associated with several forms of lymphomas, especially NHL and HL, and increased incidence of NHL has been documented among those with more serious immunodeficiency. China’s general population is known to have a high prevalence of EBV infection—one recent study found EBV prevalence among Chinese children to be 50% by age 3 and 90% by age 8 [42]. Both Zhang et al. and Chen et al. found that NHL was the most commonly observed malignancy [27,29], and Chen et al. found NHL incidence of 329 per 100,000 (SIR 23.7) for males and 256 per 100,000 (SIR 22.4) for females, whereas incidence of HL was 18 per 100,000 (SIR 9.4) for males and 14 per 100,000 (SIR 7.8) for females [29]. However, limitations in the data we collected meant we were only able to report incidence for all lymphomas as a group—SIR of 13.9 for males, 16.0 for females, and ASIRC of 63.1 per 100,000 among males and 48.0 per 100,000 among females, consistent with previous studies of NHL in the post-ART era [7,12,18,26].

HPV Infection

A broad range of cancers, mostly affecting the anogenital and head and neck areas, are known to be caused by infection with one or more of the high-risk subtypes of HPV [43]. Female PLHIV in China have a high prevalence of HPV co-infection. In one recent study, prevalence of carcinogenic HPV infection among women with HIV was nearly 40% [44]. ICC is a major health problem in China, causing the deaths of approximately 40,000 women annually. We observed 128 cases of ICC in our cohort, for an ASIRC of 47.6 per 100,000 and SIR of 3.8, consistent with previous studies in which females with HIV were estimated to be roughly 5 times more likely to develop ICC [20,21,23,24]. Zhang et al. found a higher SIR of 68.1, and all 14 cancer cases observed among women in the study were ICC [27]. However, in the Taiwan study, incidence of ICC was 413 per 100,000, with SIR at 14.0 [29].

HPV infection is also common among Chinese men who have sex with men (MSM). A recent study in three Chinese cities found high-risk subtype HPV infection among 51% of participants overall, and among 70% of participants with HIV infection [45]. HPV-associated anal cancer has risen to a steady high level among MSM globally [8]. However, very few cases of anal cancer among PLHIV in Mainland China had previously been reported. Zhang et al. found zero cases of anal cancer [27], Yang et al. found 3 among men [28], and we found only 2 (ASIRC: 0.2 per 100,000, SIR: 2.9). A possible explanation for the low incidence of anal cancer thus far observed among men with HIV infection in China is the relatively low prevalence of HIV among MSM during the study period. In our study, only 7.3% of participants reported acquiring HIV infection via homosexual contact. However, the rapid increase of HIV infection in recent years and a high prevalence of HPV infection among MSM in China predicts a rise in anal cancer cases in the future, which may approximate the incidence rate in the Taiwan cohort (45.3 per 100,000, SIR 18.5) [29].

HBV and HCV Infection

HBV and HCV infection are associated with hepatocellular carcinoma, which accounts for >90% liver cancers. Liver cancer is known to be more prevalent and have higher incidence in east and southeast Asia, with China alone having 50% of all new cases of liver cancer worldwide in 2012 [30]. Incidence of liver cancer is higher among those with HIV in the US and other European countries [5,7,11,12,17,19,23,24]. In China, HIV/HCV co-infection is very common among former blood product donors and recipients as well as people who inject drugs (PWID) [46], and the prevalence of HBV/HIV and HCV/HIV co-infection, and HBV/HCV/HIV triple infection were 8.7%, 18.2%, and 3.3%, respectively, among PLHIV in China’s NFATP [47]. As expected, a high SIR for liver cancer was found by Zhang et al. (6.0, male and female combined) [27], and 14 liver cancer cases were identified by Yang et al., which accounted for 9% of all malignancies observed [28]. In our study, liver cancer had a very high ASIRC among males, 145.7 per 100,000, similar to the Taiwan study (185 per 100,000) [29], and an SIR, 3.9 in males, 5.2 in females, similar to previous studies [5,7,11,12,17,19,23,24].

Other infections may also impact incidence of cancers in PLHIV in China. For example, Helicobacter pylori has been associated with stomach cancer. Abovementioned HPV infection has also been associated with penile cancer, oral cavity cancers, other cancers of the head and neck, and non-melanoma skin cancers, and EBV infection has been linked to nasopharyngeal cancer.[8] Incidence rates of all these malignancies were elevated in our study with the one exception—stomach cancer among male PLHIV was not elevated, which is notable since China bears nearly half of the global burden of non-cardia gastric cancers [48].

Smoking

Previous studies have found that tobacco smoking, as well as the incidence rate of lung cancer are higher among PLHIV than in the general population [3,4,19]. As anticipated, lung cancer incidence was very high in our study—ASIRC of 226.0 per 100,000 among males and 66.8 per 100,000 among females, and SIR of 4.8 for males and 4.2 for females. The incidence rate of lung cancer in our study was higher than the Taiwan cohort [29], which may be attributed to higher prevalence of tobacco smoking in males in Mainland China (53%) than in Taiwan (32–38%) [49,50].

To the authors’ knowledge the present study was the first ever nationwide investigation of the spectrum of malignancies among PLHIV in China and adds to a literature containing few such studies in middle-income country settings. The very large size and nationwide scope of our cohort were two important strengths of our study, which facilitated detection of cancers with low incidence and perhaps improved the overall accuracy of our incidence estimates.

Nevertheless, our study had several limitations. Firstly, only individuals who had been diagnosed with HIV infection could be included in our analysis, and cancer incidence rate in undiagnosed PLHIV remains unknown. Moreover, we could not evaluate cancer cases among the 10.2% of cases (45,261 of 444,712) that were excluded due to either death or loss to follow-up prior to the study’s start or no record of follow-up during the study period. However, a near 90% inclusion rate suggests that our study population is indeed nationally representative. Secondly, because original pathology findings were not included in CRIMS records, their re-examination for accuracy of cancer diagnosis could not be performed. This resulted in some cancer cases lacking a specific diagnosis (classified in our study as ill-defined/unspecified) and likely others being misdiagnosed, both of which may have introduced some miss-classification bias that could have caused under- or over-estimation of some incidence calculations. An example of this may be found in the surprisingly high numbers of brain and CNS cancers observed in our study population. It is likely that some of these cases were, in fact, primary CNS lymphomas [51]. Similarly, it is likely that some cancer cases in our study population went undiagnosed. This too would have biased our results toward under-estimation of cancer incidences. Thirdly, due to the lack of specificity in nearly half of the lymphoma cases in the cohort, we had to classify all lymphomas together into a single group. This prevented comparison of ADC and NADC incidence within the current study and previous studies.