The increasing survival of HIV-infected individuals with access to highly active antiretroviral therapy (HAART) has led to dramatic changes in mortality patterns of people with HIV/AIDS.1,2 Although mortality rates for AIDS-related conditions have been decreasing over time, those of non–AIDS-related ones remained unchanged or decreased very slightly.3–5 In particular, non–AIDS-defining cancers (non-ADCs, ie, cancers other than Kaposi sarcoma, non–Hodgkin lymphoma, and cervix uteri) have become the leading non–AIDS-related causes of death in the late HAART era, among both people with HIV1,3 and people with AIDS (PWA).6
Nonetheless, the risk of death as compared with the HIV-uninfected population is still far from being assessed for each cancer site/type in the late HAART era, as only few studies have quantified excess mortality for specific cancers among people with HIV4 or PWA.6 In addition, most mortality studies were solely based on the underlying cause of death (ie, the disease or injury which initiated the train of morbid events leading directly to death),7 which is poorly informative of the complexity of the morbidity conditions that affect people with HIV/AIDS. Furthermore, such approach does not allow a direct comparison with the uninfected population because most death certificates of people with HIV/AIDS report “HIV/AIDS” as the underlying cause of death.6,8 To overcome such limits, several methods have been used.3–6,8,9 For instance, in a previous Italian investigation aimed at quantifying the excess cancer mortality among PWA, as compared with the general population, we reassigned the underlying cause of death by excluding HIV/AIDS reference in the death certificates.10,11
Currently, the availability in Italy of multiple-cause-of-death (MCoD) data (ie, all the conditions reported in death certificates, beyond the underlying cause of death) allows making a direct comparison of the death certificates of people with HIV/AIDS with those of people without. Furthermore, it allows investigating in detail the presence at death of specific cancer sites/types and other diseases even if these conditions were not selected as the underlying cause. Using MCoD data, this nationwide population-based study aimed to assess excess risk of cancer-related deaths among Italian PWA, as compared with people without AIDS (non-PWA), in the late HAART era.
This retrospective cohort study is part of a larger epidemiological, population-based, nationwide investigation on the survival and mortality patterns of PWA in Italy.12 The following data sources were used: (1) the National AIDS Registry (RAIDS) at the National Health Institute, which collects mandatory data on all people newly diagnosed with AIDS (according to the 1993 revised European definition13) and (2) the National Register of Causes of Death (RCoD) at the National Institute of Statistics, which collects all death certificates mandatorily issued in Italy.14 MCoD data are coded by RCoD, according to the International Classification of Diseases 10th revision (ICD-10) rules and provisions issued by the World Health Organization.7
To compare the causes of death between PWA and non-PWA, RAIDS and RCoD data were linked for the concurrently available period, ie, 2006–2011. The record-linkage procedure was conducted in observance of current laws regulating the use of RAIDS and RCoD data (ie, inclusion of the investigation in the Italian National Statistical Plan according to permission of the Data Protection Authority).12 The record-linkage procedure was carried out by means of a validated, semi-automated, software application using names, surnames, and dates of birth that were blinded to the operator and removed from the output.15,16 To guarantee the highest completeness of the record-linkage procedure, we excluded: (1) PWA who were residing in the provinces of Trento and Bolzano, because their names were not available in RCoD; and (2) PWA who were foreign citizens, because of possible bias deriving from a higher frequency of spelling errors in their names when recorded at RAIDS and from a higher propensity of these people to migrate abroad and of being lost to follow-up.
For the aims of this study, “PWA deaths” were identified by linked records in RAIDS and RCoD, whereas “non-PWA deaths” were identified by RCoD records not linked with RAIDS. Furthermore, to avoid the inclusion of HIV-infected people in the comparison group, we excluded from the “non-PWA deaths” category those records reporting AIDS/HIV-related conditions (ie, ICD-10 codes B20–B24).
To improve comparability between the study groups, the present analysis was restricted to PWA aged 15–74 years at diagnosis or at death. Furthermore, we excluded PWA with <1 month of follow-up, as the day (within the date) of AIDS diagnosis was often missing; this selection allowed also the exclusion of those PWA diagnosed at death. Our analysis finally included 1229 deaths occurred among 5285 PWA and 952,019 non-PWA deaths.
Person-years at risk of death for PWA were calculated from the date of AIDS diagnosis to the date of death, or to December 31, 2011, whichever came first. PWA who had been diagnosed within 74 years of age but who died thereafter were censored at their 75th birthday. The risk of death of PWA, as compared with non-PWA, was estimated using sex- and age-standardized mortality ratios (SMRs).17 SMRs were calculated as the ratio between the observed number of deceased PWA who had a specific cancer to the expected one, estimated on the basis of sex- and age-specific (quinquennia) mortality rates among non-PWA, multiplied by the person-years at risk among PWA. Mortality rates for non-PWA were computed using, as numerator, the observed number of deceased non-PWA for that specific condition, and, as denominator, the average resident Italian population of same sex and age in the study period (after the exclusion of Trento and Bolzano provinces and foreign citizens), as a proxy of person-years at risk of death. Corresponding 95% confidence intervals (CIs) were computed using the exact Poisson method as appropriate for rare events (ie, ≤5).18 However, cancer sites with a total number of observed deaths <4 were included in the auxiliary Table A (see Supplemental Digital Content, http://links.lww.com/QAI/A820).
Secondary malignancies and malignant neoplasms not otherwise specified (ie, ICD-10 codes: C77–C80) were not considered in these analyses, as they were almost always (except in 8 cases out of 141 PWA deaths) reported in death certificates together with defined cancers and were, therefore, not relevant to the study aim.
Table 1 shows the distribution of 5285 Italian PWA diagnosed during 2006–2011 and the 1229 PWA who had died as of December 31, 2011, according to main characteristics at AIDS diagnosis. Mean follow-up time was 2.5 years, for a total of 14,180 person-years at risk of death. The proportion of observed deaths among PWA increased with age at AIDS diagnosis, whereas it decreased with increasing years of education; moreover, it was particularly high among PWA infected through injecting drug use (IDU) (31%), and it was the lowest among homosexual PWA (17%).
Table 2 shows the distribution of malignant neoplasms reported in death certificates of PWA and non-PWA, using MCoD data. AIDS-defining cancers (ADCs) were reported in 22.9% of death certificates of PWA: non–Hodgkin lymphoma in 18.0%, Kaposi sarcoma in 5.1%, and cervical cancer in 2.5%. Similar figures were found among PWA aged 15–49 and 50–74 years, with the exception of no mention of cervical cancer among female PWA aged ≥50 years. Conversely, these same malignancies were rarely reported as causes of death in non-PWA (<2% of total deaths, in all age groups).
Non-ADCs were reported in 10.3% of total deaths among PWA. The most frequent type was lung cancer (in 3.1% of death certificate, 38 deaths), followed by liver cancer (1.4%, 17 deaths), and Hodgkin lymphoma (1.0%, 12 deaths). Among female PWA, unspecified uterine cancer was reported in 1.7% of death certificates (4 cases). Non-ADCs were more frequently observed in PWA who died between the ages of 50 and 74 years (13.8% of death certificates) than in younger ones (7.9%). As expected, non-ADCs were by far more frequently observed among non-PWA (46.8% of death certificates).
Tables 3 and 4 show the observed and the expected numbers of cancer-related deaths with the corresponding SMRs, overall and in the selected sub-groups. ADCs, in particular Kaposi sarcoma, showed extremely high excess mortality, especially among PWA <50 years (Table 3). When considering all non-ADCs together, a 7.3-fold excess risk of death was observed (95% CI: 6.1 to 8.7) with a higher excess in the youngest group of PWA (SMR = 14.2, 95% CI: 10.8 to 18.5, for 15–49 years vs. 5.3, 95% CI: 4.1 to 6.7, for 50–74 years). Remarkably elevated excess risks of death emerged for anal cancer (SMR = 228) and Hodgkin lymphoma (SMR = 122). Statistically significant excess mortality was also found for not otherwise specified uterine cancers (SMR = 52.5, 95% CI: 14.3 to 135), liver (SMR = 13.2, 95% CI: 7.7 to 21.1), skin melanoma (SMR = 10.9, 95% CI: 3.0 to 27.8), lung (SMR = 8.0, 95% CI: 5.7 to 11.0), head and neck (SMR = 7.8, 95% CI: 3.6 to 14.9), leukemia (SMR = 7.6, 95% CI: 2.4 to 17.7), and colon-rectum (SMR = 5.4, 95% CI: 2.6 to 10.0). The SMRs were generally higher in the 15- to 49-years age group (Table 3). In particular, elevated excess risks among younger PWA emerged for liver (SMR = 38.8 among PWA who died at ages 15–49 years vs. SMR = 7.6 at ages 50–74 years) and lung cancers (SMR = 22.0 and 5.7, respectively). The sex-stratified analysis (see Table B, Supplemental Digital Content, http://links.lww.com/QAI/A820) showed similar SMRs for both ADCs (SMR = 424, 95% CI: 372 to 482, in men; SMR = 487, 95% CI: 358 to 647 in women) and non-ADCs (SMR = 7.1, 95% CI: 5.8 to 8.6, in men; SMR = 9.1, 95% CI: 5.6 to 14.1 in women). Female PWA reported higher excess mortality than male PWA for non–Hodgkin lymphoma (SMR = 339, 95% CI: 292 to 392, in men; SMR = 554, 95% CI: 386 to 771 in women) and lung cancer (SMR = 7.4, 95% CI: 5.1 to 10.4 in men; SMR = 17.3, 95% CI: 5.6 to 40.3 in women).
Both PWA infected through IDU and through sexual intercourses reported extremely elevated excess mortality for ADCs (Table 4). PWA who had been infected through IDU showed particularly high excess mortality for non-ADCs, the overall SMR being equal to 20.0 (95% CI: 15.1 to 26.1). Of note, 14 out of 17 (82%) liver cancers occurred among IDU PWA, corresponding to an SMR of 74.7 (95% CI: 40.8 to 125.3). Also the SMRs of anus (SMR = 440.4, 95% CI: 53.3 to 1591), not otherwise specified uterus (SMR = 156.8, 95% CI: 32.3 to 458.2), lung (SMR = 29.3, 95% CI: 17.4 to 46.3), and head and neck cancers (SMR = 23.6; 95% CI: 7.6 to 55.0) were extremely elevated in this group.
Considering PWA infected through sexual intercourses, the SMR for non-ADCs overall was 4.9 (95% CI: 3.8 to 6.2). Very high SMRs emerged especially for anal cancer (SMR = 189.5; 95% CI: 39.1 to 553.9), Hodgkin lymphoma (SMR = 115.9; 95% CI: 50.0 to 228.4), but significant excess mortality was detected also for leukemias (SMR = 8.2; 95% CI: 2.2 to 21.0), lung (SMR = 5.1; 95% CI: 3.1 to 8.0), and colorectal cancer (SMR = 3.5; 95% CI: 1.1 to 8.3).
This study provided nationwide, population-based estimates of the excess cancer mortality among Italian PWA, as compared with people without HIV/AIDS, in the late-HAART era. In addition to extremely elevated risks of death for ADCs, the study documented a statistically significant higher mortality for cancers not directly associated with HIV/AIDS, with an overall 7.3-fold excess risk for all non-ADCs combined. This latter figure can be interpreted as a combination of the increased incidence of such malignancies15,19–22 and of the reduced cancer survival of people with HIV/AIDS.16,23,24
Overall, significant excess mortality emerged for cancers associated with viruses, for which HIV-infected individuals are likely to lose the immune control of infections, and for cancers associated with unhealthy behaviors, such as tobacco smoking. All these risk factors have been shown to be more common among HIV-infected individuals than among uninfected ones.1,4,5,25,26 Furthermore, mortality in HIV patients without such risk factors was found to be very similar to that of the non–HIV-infected individuals in a Danish cohort.26 In particular, the poorer survival observed among HIV-infected people via IDU was recently reported to be mostly due to hepatitis C virus coinfection.27 In our study, PWA infected via IDU also reported very high SMRs.
As expected, extremely high SMRs were documented for the 3 ADCs, which were strongly associated with viruses.28 A very high excess risk emerged also for the group of not otherwise specified uterine cancers, where several misclassified cervical cancers are included.29
Among non-ADCs, very high excess risks were registered for Hodgkin lymphoma (associated to EBV28) and anal carcinoma (associated to HPV28), 2 cancers that deserve particular attention as the elevated corresponding SMRs were because of the combination of their relative paucity in the general population and to their increased incidence among HIV-infected people.15,19,22 Of note, 2 out of 5 PWA death certificates reporting anal cancer mentioned also Kaposi sarcoma, in line with findings of a large prospective cohort of HIV-positive individuals.21 This cancer was also found to be more frequent among homosexual PWA19 (3 out of 5 cases, in our study). Liver cancer, which is strongly associated to infections with hepatitis C and B viruses,28 showed a particularly high SMR, especially among IDU PWA, similarly to other studies.19 This tumor was reported together with liver cirrhosis (ICD-10 code K74.6) in 10 out of 17 death certificates (59%) and with chronic viral hepatitis C (ICD-10 code B18.2) in 5 cases (29%). Highly significant excess risks of death were also seen for lung and head and neck cancers—associated with tobacco smoking28—in particular among PWA infected through IDU but also among PWA who died at age 50–74 years (an age period in which lung cancer is very common also among the general population14). On the other hand, colorectal cancer, leukemia, and skin melanoma—which also reported significant SMRs—were not generally found to be at increased incidence risk among HIV/AIDS patients.15,19 The excess death risk observed for these cancers can be explained by a poorer survival of HIV/AIDS patients after cancer onset.15,23 Some misclassification of anal cancer into rectal cancer in death certificates cannot be excluded (SMR = 19.0; 95% CI: 7.6 to 39.1 for rectal cancer, see Table A, Supplemental Digital Content, http://links.lww.com/QAI/A820).
The spectrum of non-ADCs, for which we found excess mortality, and the magnitude of SMRs are in line with our previous investigation11 and with other studies conducted in high-income countries.1,4,6,20 The proportion of deaths associated with non-ADCs was higher in the present study, referring to the period 2006–2011, than in the previous one conducted in the period 1999–200611 (10.3% vs. 7.4%). This difference did not seem attributable to the use of MCoD: among the 17 death certificates of PWA that reported more than one cancer, the majority were ADCs. The increasing proportion of non-ADCs at death is in agreement with findings of other investigations.1,3,6 Indeed, this proportion increased from 9% in 1999–2000 to 23% in 2009–2011 among HIV-positive individuals under treatment in a large multi-cohort collaboration3 and from 7% to 16.9% between 1996–1997 and 2006–2011 among PWA in San Francisco.6
Among the main strengths of this study is the use of MCoD data, which allowed direct comparisons with non-PWA for all the conditions contributing to death without the need to resort to a manual review of death certificates, as customary in earlier studies,10,11 thus, improving reproducibility. Comparisons based on the underlying cause may be affected by poor detail and would prevent the identification of specific conditions.6,8 Indeed, in our data, 73% (293 cases) of PWA reporting a cancer in death certificate had HIV/AIDS codes as the underlying cause of death (ie, ICD-10 codes B20–B24)—a value similar to that observed in other investigations,8 often with a poor level of detail (eg, 50% had ICD-10 code B22.7 = “HIV disease resulting in multiple diseases,” 7% had unspecific cancer codes B21.7–21.8).
Issues of reliability regarding causes of death derived from death certificate statistics are well known,30 including a possible lack of specificity and/or underreporting because of the limited knowledge of the certifying physician regarding the medical history of the deceased. This could have differently affected the compilation of death certificates of people with or without HIV infection. Nonetheless, in our investigation, the same coding rules were applied to both study groups, thus limiting information bias.
It is worth remembering that our data included only HIV-infected individuals having already had an AIDS diagnosis. Thus, study results cannot be referred to HIV-infected people at an earlier stage of immunodeficiency. In particular, given that AIDS diagnosis can be a consequence of the diagnosis of an ADC, SMRs for ADCs could be much lower in the case of HIV-infected people without AIDS (in our data, approximately 84% of PWA reporting ADCs at death reported also ADCs at diagnosis).
The limited number of person-years at risk, resulting in a low number of observed deaths for specific cancer types, was also a study limitation. However, given that most of the PWA deaths occurred within the first 6 months after AIDS diagnosis,12 the median follow-up of 2.5 years was sufficiently long for observing most of the events of interest. Nonetheless, SMRs for very a low number of deaths should be interpreted with caution.
Completeness was the main strength of this investigation. The full coverage of the Italian population by the 2 used data sources allowed robust comparisons between PWA and the general population. Furthermore, the high sensitivity of the record-linkage procedure15,16 allowed to keep the number of losses to follow-up very limited.
In conclusion, findings from our study stress the need of monitoring the burden of both virus-related and non–virus-related cancers among PWA in the late HAART era, as they are still at a higher risk of death than the HIV-negative population. This applies in particular to PWA who acquired HIV infection through IDU. Indeed, our results call for taking primary and secondary preventive actions to reduce both cancer incidence and mortality among people with HIV or AIDS.
The authors thank Mrs. Luigina Mei for editorial assistance and Mr. Stefano Boros for RAIDS data management. The authors gratefully acknowledge the contribution of Dr. Paolo De Paoli, the Scientific Director of IRCCS CRO-Aviano, for supporting the original project (Istituto Superiore di Sanità-Progetto Nazionale AIDS 2006, ISS 20G.3 and 20 G.12), from which this research was derived.
1. Vandenhende MA, Roussillon C, Henard S, et al. Cancer-Related causes of death among HIV-infected patients in France in 2010: Evolution since 2000. PLoS One. 2015;10:e0129550.
2. Antiretroviral Therapy Cohort Collaboration. Life expectancy of individuals on combination antiretroviral therapy in high-income countries: a collaborative analysis of 14 cohort studies. Lancet. 2008;372:293–299.
3. Smith CJ, Ryom L, Weber R, et al. Trends in underlying causes of death in people with HIV from 1999 to 2011 (D:A:D): a multicohort collaboration. Lancet. 2014;384:241–248.
4. Helleberg M, Kronborg G, Larsen CS, et al. Causes of death among Danish HIV patients compared with population controls in the period 1995–2008. Infection. 2012;40:627–634.
5. Antiretroviral Therapy Cohort Collaboration. Causes of death in HIV-1-infected patients treated with antiretroviral therapy, 1996–2006: collaborative analysis of 13 HIV cohort studies. Clin Infect Dis. 2010;50:1387–1396.
6. Schwarcz SK, Vu A, Hsu LC, et al. Changes in causes of death among persons with AIDS: San Francisco, California, 1996–2011. AIDS Patients Care STDS. 2014;28:517–523.
7. World Health Organization, WHO. International Statistical Classification of Diseases and Related Health Problems. Tenth Revision (ICD-10). Geneva, Switzerland: WHO; 1992. Available at http://www.who.int/classifications/icd/en/
8. Whiteside YO, Selik R, An Q, et al. Comparison of rates of death having any death-certificate mention of heart, kidney or liver disease among persons diagnosed with HIV infection with those in the general US population, 2009–2011. Open AIDS J. 2015;9:14–22.
9. Simard EP, Pfeiffer RM, Engels EA. Mortality due to cancer among people with AIDS: a novel approach using registry-linkage data and population attributable risk methods. AIDS. 2012;26:1311–1318.
10. Serraino D, De Paoli A, Zucchetto A, et al. The impact of Kaposi sarcoma and non-Hodgkin lymphoma on mortality of people with AIDS in the post-HAART era. Cancer Epidemiol. 2010;34:257–261.
11. Zucchetto A, Suligoi B, De Paoli A, et al. Excess mortality
for non-AIDS-defining cancers among people with AIDS. Clin Infect Dis. 2010;51:1099–1101.
12. Suligoi B, Zucchetto A, Grande E, et al. Risk factors for early mortality after AIDS in the cART era: a population-based cohort study in Italy. BMC Infect Dis. 2015;15:229.
13. Ancelle-Park R. Expanded European AIDS case definition. Lancet. 1993;341: 441.
15. Dal Maso L, Polesel J, Serraino D, et al. Pattern of cancer risk in persons with AIDS in Italy in the HAART era. Br J Cancer. 2009;100:840–847.
16. Dal Maso L, Suligoi B, Franceschi S, et al. Survival after cancer in Italian persons with AIDS, 1986–2005: a population-based estimation. J Acquir Immune Defic Syndr. 2014;66:428–435.
17. Breslow NE, Day NE. Statistical Methods in Cancer Research, Volume II. The Design and Analysis of Cohort Studies. IARC Scientific Publications. Lyon, France: International Agency for Research on Cancer; 1987.
18. Ulm K. A simple method to calculate the confidence interval of a standardized mortality ratio
(SMR) Am J Epidemiol. 1990;131:373–375.
19. Robbins HA, Pfeiffer RM, Shiels MS, et al. Excess cancers among HIV-infected people in the United States. J Natl Cancer Inst. 2015;107.
20. Raffetti E, Albini L, Gotti D, et al. Cancer incidence and mortality for all causes in HIV-infected patients over a quarter century: a multicentre cohort study. BMC Public Health. 2015;15:235.
21. Worm SW, Bower M, Reiss P, et al. Non-AIDS defining cancers in the D:A:D Study–time trends and predictors of survival: a cohort study. BMC Infect Dis. 2013;13:471.
22. Shiels MS, Cole SR, Kirk GD, et al. A Meta-analysis of the incidence of non-AIDS Cancer in HIV-infected individuals. J Acquir Immune Defic Syndr. 2009;52:611–622.
23. Coghill AE, Shiels MS, Suneja G, et al. Elevated cancer-specific mortality among HIV-infected patients in the United States. J Clin Oncol. 2015;33:2376–2383.
24. Gotti D, Raffetti E, Albini L, et al. Survival in HIV-infected patients after a cancer diagnosis in the cART era: results of an Italian Multicenter Study. Plos One. 2014;9:e94768.
25. Wada N, Jacobson LP, Cohen M, et al. Cause-specific mortality among HIV-infected individuals, by CD4+ cell count at HAART initiation, compared with HIV-uninfected individuals. AIDS. 2014;28:257–265.
26. Obel N, Omland LH, Kronborg G, et al. Impact of non-HIV and HIV risk factors on survival in HIV-infected patients on HAART: a population-based nationwide cohort study. PLoS One. 2011;6:e22698.
27. May MT, Justice AC, Birnie K, et al. Injection drug use and Hepatitis C as risk factors for mortality in HIV-infected individuals: the Antiretroviral Therapy Cohort Collaboration. J Acquir Immune Defic Syndr. 2015;69:348–354.
28. IARC Working group on the Evaluation of Carcinogenic risks in Humans. Biological agents.Volume100B. A review of human carcinogenesis. IARC Monogr Eval Carcinog Risks Hum. 2012;100:1–441.
29. Loos AH, Bray F, McCarron P, et al. Sheep and goats: separating cervix and corpus uteri from imprecisely coded uterine cancer deaths, for studies of geographical and temporal variations in mortality. Eur J Cancer. 2004;40:2794–2803.
30. Grippo F, Grande E, Pace M. Cause-of-death reporting: how to measure inaccurate completion. GENUS. 2013;69:25–45.
HIV/AIDS; multiple causes of death; excess mortality; standardized mortality ratio
Supplemental Digital Content
Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved.