Reducing rates of HIV-related cancer in US from 1996-2012

Simon Collins, HIV i-Base

A large registry-linked database of almost 450,000 people living with HIV in eight US states and Puerto Rico reported significant reductions in the risk of many cancers over time.

These results, published in the November 2017 edition of the Lancet HIV, were strongly linked to availability of earlier and better HIV treatment. Importantly, there were no cancers where risks increased over time. 

From 1996 to 2012, the HIV/AIDS Cancer Match (HACM) Study included 21,294 newly diagnosed cancers, from more than three million person-years of follow-up (PYFU). However, each registry contributed data for different periods: Colorado (1996–2007), Connecticut (2005–10), Georgia (2004–12), Maryland (2008–12), Michigan (1996–2010), New Jersey (1996–2012), New York (2001–12), Puerto Rico (2003–12), and Texas (1999–2009).

Standardised incidence ratios (SIR) were compared to data from the US general population (which would also include HIV positive people), adjusted for sex, age, race or ethnic group, calendar year, and registry. The analysis also compared time with AIDS to only HIV, and used four time periods:  1996–99, 2000–04, 2005–08, and 2009–12. The majority of follow-up cover the last two time periods (each with > 1 million PYFU). Participants in the later periods were also more likely to contribute >10 years of PYFU.

Overall, rates of cancer were approximately 70% higher for HIV positive people (SIR: 1.69, 95%CI: 1·67–1·72, p<0.0001).

Rates of HIV-related cancers (14-fold higher) or those linked to other viruses (5-fold higher) were still significantly higher in HIV positive people compared to rates in the general population (p<0.0001). Rates were also higher for some non-virus-related cancers (notably lung, non-HPV oral), but significantly lower for others (including colon, stomach, female breast, prostate. kidney/renal pelvis). In the cases of lower rates, it is unclear whether, for example, this might be associated with more monitoring in the general population (that might not be accessed by HIV positive people) or a survival bias linked to HIV positive people being vulnerable to other cancers at younger age. For many other cancers, rates were similar, irrespective of HIV status. See Table 1.

There were also a few examples of where the trends of some cancers to reduce over time were not significant: Burkitt’s lymphoma (p-0.8944), cervical cancer (p=0.0989) and Hodgkin’s lymphoma (p=0.862).

However, the results were also limited by not having access to individual data for HIV positive people, for example for CD4, viral load or ART used etc.

Although the paper comments that earlier ART and wider access to ART might continue to reduce cancer rates – both of which have significantly changed in the five years since the cut-off for data collected in this study – it also notes that ART might not fully reverse the effects of immunosuppression, especially for people who are diagnosed late or who delay ART. Given many cancers take decades to develop, the limited data on HIV and cancer in older people makes continued monitoring vital.

Table 1: Incidence of selected cancers in HIV positive vs US general population, 1996-2012
Cancer type n SIR 95% CI p
Overall 21,294 1·69 1·67–1·72 <0.0001
AIDS-defining cancers 6,384 13·97 13·63–14·32 <0.0001
Kaposi’s sarcoma 2269 498·11 477·82-519·03 <0.0001
Non-Hodgkin lymphoma 3687 11·51 11·14-11·89 <0.0001
Cervix 428 3·24 2·94-3·56 <0.0001
Other virus-related cancers 4144 5·39 5·23-5·55 <0.0001
Anal 1568 19·06 18·13-20·03 <0.0001
Liver 1104 3·21 3·02-3·41 <0.0001
Hodgkin’s lymphoma 875 7·70 7·20-8·23 <0.0001
Virus-unrelated 10,200 0·92 0·90-0·94 <0.0001
Lung 2475 1·97 1·89-2·05 <0.0001
Oral (non-HPV) 343 2·20 1·98-2·45 <0.0001
Colon* 477 0·61 0·56-0·67 <0.0001
Stomach* 185 0·74 0·64-0·86 <0.0001
Female breast* 688 0·63 0·58-0·68 <0.0001
Prostate* 1522 0.48 0·46-0·51 <0.0001
Kidney/renal pelvis* 360 0.74 0·66-0·82 <0.0001

* Examples of significantly reduced rates in HIV positive people

SIR: standardised incidence ratio; CI: confidence interval


These results, although retrospective and only based on US data to 2012, are important for the size of the cohort and are generally very positive.

Higher rates of smoking-related cancers emphasise the importance of support for lifestyle changes including smoking cessation.


Hernández-Ramírez RU et al. Cancer risk in HIV-infected people in the USA from 1996 to 2012: a population-based, registry-linkage study. Lancet HIV, 4(11)e495–e504. doi: 10.1016/S2352-3018(17)30125-X. (November 2017).

Links to other websites are current at date of posting but not maintained.