Background: It is unknown if the carcinogenic effect of smoking is influenced by CD4 þ cell count and viral load in persons living with HIV.Material and methods: RESPOND participants with known smoking status were included. Poisson regression adjusting for baseline confounders investigated the interaction between current CD4 þ /viral load strata [good (CD4 þ cell count !500 cells/ml and viral load <200 copies/ml], poor [CD4 þ cell count 350 cells/ml and viral load >200 copies/ml] and intermediate [all other combinations]), smoking status and all cancers, non-AIDS defining cancers (NADCs), smoking-related cancers (SRCs) and infection-related cancers (IRCs).Results: Out of 19 602 persons, 41.3% were never smokers, 44.4% current and 14.4% previous smokers at baseline. CD4 þ /viral load strata were poor in 3.4%, intermediate in 44.8% and good in 51.8%. There were 513 incident cancers; incidence rate 6.9/1000 person-years of follow-up (PYFU) [95% confidence interval (95% CI) 6.3-7.5]. Current smokers had higher incidence of all cancer (adjusted incidence rate ratio 1.45; 1.17-1.79), NADC (1.65; 1.31-2.09), SRC (2.21; 1.53-3.20) and IRC (1.38; 0.97-1.96) vs. never smokers. Those with poor CD4 þ /viral load had increased incidence of all cancer (5.36; 95% CI 3.71-7.75), NADC (3.14;, SRC (1.82; 0.76-4.41) and IRC (10.21; 6.06-17.20) vs. those with good CD4 þ /viral load. There was no evidence that the association between smoking and cancer subtypes differed depending on the CD4 þ /viral load strata (P > 0.1, test for interaction).
Conclusion:In the large RESPOND consortium, the impact of smoking on cancer was clear and reducing smoking rates should remain a priority. The association between current immune deficiency, virological control and cancer was similar for never smokers, current smokers and previous smokers suggesting similar carcinogenic effects of smoking regardless of CD4 þ cell count and viral load.