HIV Articles  
Back 
 
 
Excess Mortality for Non-AIDS-Defining Cancers among People with AIDS
 
 
  Clinical Infectious Diseases Nov 12010
 
"In conclusion, this analysis provides estimates of the excess risk of death for NADCs among people with AIDS in the HAART era. Our findings also highlight the importance of monitoring the cancer burden on mortality of people with AIDS, offering specialists in infectious diseases and oncologists an additional means of evaluation of the death threat of NADCs to people with AIDS."
 
"In this analysis of the risk of death for NADCs among Italian people with AIDS in the HAART era, we found a nearly 7-fold excess, compared with the general population of the same sex and age. Similar comparative studies were conducted in the pre-HAART era, when mortality due to AIDS-defining events had a much stronger impact and NADCs were less common [6, 9]. One investigation, which used data from the US population from 1990 through 1995, found an about 2-fold higher relative risk of death with any mention of NADCs among HIV-infected people aged 25-44 years, compared with the uninfected population"
 
"Liver cancer was among the most common causes of death (12% of NADCs), especially among injection drug users, supporting a key role of hepatitis virus infection on mortality in people with AIDS. Hodgkin lymphoma, which is associated to Epstein-Barr virus, had also a strong impact on death (12%), with a very high SMR. This finding can be partly explained by the knowledge that, among HIV-infected individuals, Hodgkin lymphoma is more common and bears a worse prognosis than among the general population [13]. Anal cancer-a human papilloma virus-associated tumor-demonstrated an extremely elevated SMR (ie, 270), an excess that reflects the increased incidence of such illness in people with HIV infection or AIDS [1, 2] and its rarity as cause of death in the general population of the same age of people with AIDS."
 
"We emphasize that our findings refer only to people with AIDS, who may not be representative of HIV-infected people. Because HAART has decreased the incidence of AIDS among HIV-infected individuals, such persons may die of an NADC without developing an AIDS-defining condition. Therefore, our SMRs could be potentially biased with respect to the overall population of HIV-infected people."

 
Antonella Zucchetto,1 Barbara Suligoi,3 Angela De Paoli,1 Simona Pennazza,4 Jerry Polesel,1 Silvia Bruzzone,4 Giovanni Rezza,3 Paolo De Paoli,2 Luigino Dal Maso,1,5 and Diego Serraino1
 
1Unit of Epidemiology and Biostatistics and 2Scientific Directorate, Centro di Riferimento Oncologico, Istituto di Ricovero e Cura a Carattere Scientifico, Aviano, 3Dipartimento di Malattie Infettive, Centro Operativo AIDS, Istituto Superiore di Sanita, Roma, Italy; 4Direzione Centrale per le Statistiche e le Indagini Sulle Istituzioni Sociali, Servizio Sanita e Assistenza, National Institute of Statistics, Rome, and 5Statistica Medica e Biometria, Universita degli Studi di Milano, Milan, Italy
 
During the period 1999-2006, non-AIDS-defining cancers accounted for 7.4% of deaths among Italian people with AIDS. The risk of death was 6.6-fold higher than in the general population, being particularly elevated for virus-related cancers. The study findings highlighted the importance of monitoring the cancer burden on mortality for people with AIDS.
 
Survival improvements following the availability of highly active antiretroviral therapy (HAART) have led to increased frequencies of non-AIDS-related conditions, including non-AIDS-defining cancers (NADCs; ie, cancers other than Kaposi sarcoma, non-Hodgkin lymphoma, and cervix uteri) in human immunodeficiency virus (HIV)-infected people. A large number of worldwide studies have quantified the risk of developing NADCs in the HAART era [1-5]. Among the studies investigating the causes of death in people with HIV infection, including the burden of NADCs on mortality, few have performed comparisons with sex- and age-matched people from the general population [6-9].
 
This study is part of a larger, ongoing epidemiological investigation on survival and perimortal conditions of people with AIDS in the HAART era based on mortality data from the entire, general Italian population. In previous articles, we analyzed the survival of people with AIDS and the prognostic role of AIDS-defining illnesses at diagnosis [10], as well as the impact of Kaposi sarcoma and non-Hodgkin lymphoma on mortality among people with AIDS [11]. In this article, we assess the risk of death due to NADCs among Italian people with AIDS, compared with persons in the general population of the same sex and age.
 
Methods.
 
The study design was described elsewhere [10, 11]. In brief, during the period 1999-2006, a total of 13,485 people in Italy received a diagnosis of AIDS, according to the 1993 revised European definition, and were mandatorily reported to the National AIDS Registry. The present study was restricted to 10,392 people with AIDS who were Italian citizens, who were aged >15 years, whose diagnosis was not made at autopsy, and who resided in provinces with complete death certificate information. Baseline characteristics of eligible and noneligible Italian people with AIDS were similar.
 
Vital status was updated through a record linkage with the mortality database at the National Institute of Statistics (ISTAT) by a validated software application [2, 3, 10], which identified 3209 deaths. Death certificates were reviewed at ISTAT by the same team of professionals who code the causes of death in the Italian general population-in accordance with The International Classification of Diseases, Tenth Revision (ICD-10)-to identify the underlying cause of death (ie, the disease that initiated the sequence of morbid events leading to death). Because the use of ICD-10 codes B20-B24 is restricted to people for whom HIV/AIDS is mentioned in the death certificate, we could not apply these codes in comparing the underlying causes of death of deceased people with AIDS with those of persons in the general population. To overcome such limit, the underlying causes of death of people with AIDS were coded using the same ICD-10 codes applied to the general population (ie, to those without HIV/AIDS).
 
Time at risk was calculated from date of AIDS diagnosis to the date of death (or to 31 December 2006 for living people with AIDS). The risk of death was estimated using standardized mortality ratios (SMRs) and corresponding 95% confidence intervals (CIs) [12]. For each specific NADC, the SMR was calculated as the ratio of the observed number of deaths for that NADC to the expected number of deaths based on sex- and age-specific mortality rates for the same cause in the general population.
 
Results.
 
The 10,392 Italian people with AIDS included in the study had a median duration of follow-up of 37 months (interquartile range, 12-65 months), totaling 35,224 person-years. The 3209 deceased people with AIDS had a median age at death of 42 years (interquartile range, 38-50 years); 79.1% were men, and 50.7% were injection drug users.
 
NADCs were reported in 363 death certificates (11.3%) for the 3209 deceased people with AIDS, and these cancers were the underlying cause of death for 236 (7.4%) of the patients (Table 1). Table 1 illustrates the observed and expected numbers of deaths due to the most common NADCs, with corresponding SMRs. The median time from AIDS diagnosis to death for all NADCs was 9 months, whereas the median age at death for these 236 people with AIDS was 46 years. Overall, a 6.6-fold elevated risk of death for all NADCs was observed (95% CI, 5.8-7.5), with the most elevated SMRs seen for anal cancer (SMR, 270) and Hodgkin lymphoma (SMR, 174). Significantly elevated SMRs were also recorded for cancers of the liver (SMR, 11.1), brain and central nervous system (SMR, 10.0), head and neck (SMR, 8.2), and lung (SMR, 5.9); for myeloma and leukemia (SMR, 5.9); and for cancer of the stomach (SMR, 3.1) (Table 1).
 
Table 1. Observed and Expected Numbers of Deaths and Corresponding Standardized Mortality Ratios (SMRs) for Non-AIDS-Defining Cancers (NADCs) among 10,392 People with AIDS (35,224 Person-Years), Italy, 1999-2006
 

The risks of death for NADCs for people with AIDS, compared with the general population, were higher among injection drug users (SMR, 15.5; 95% CI, 12.5-18.9) than among those in other HIV transmission categories (SMR, 4.8; 95% CI, 4.0-5.6), notably for liver cancer (SMR, 65.2; 95% CI, 40.8-98.8). The SMRs for all NADCs were also higher among people with AIDS aged <45 years (SMR, 13.7; 95% CI, 11.3-16.4) than among those aged >45 years (SMR, 4.2; 95% CI, 3.5-5.0) and among women (SMR, 9.5; 95% CI, 6.8-12.8) than among men (SMR, 6.2; 95% CI, 5.4-7.1) (data not shown).
 
Discussion.
 
In this analysis of the risk of death for NADCs among Italian people with AIDS in the HAART era, we found a nearly 7-fold excess, compared with the general population of the same sex and age. Similar comparative studies were conducted in the pre-HAART era, when mortality due to AIDS-defining events had a much stronger impact and NADCs were less common [6, 9]. One investigation, which used data from the US population from 1990 through 1995, found an about 2-fold higher relative risk of death with any mention of NADCs among HIV-infected people aged 25-44 years, compared with the uninfected population
[6]. Another work, conducted during 1994-1998 in San Francisco, California, found an overall SMR of 4.1 for NADCs among people with AIDS [9]-a risk compatible with our estimate.
 
We observed particularly elevated SMRs for cancers with viral etiologies. Liver cancer was among the most common causes of death (12% of NADCs), especially among injection drug users, supporting a key role of hepatitis virus infection on mortality in people with AIDS. Hodgkin lymphoma, which is associated to Epstein-Barr virus, had also a strong impact on death (12%), with a very high SMR. This finding can be partly explained by the knowledge that, among HIV-infected individuals, Hodgkin lymphoma is more common and bears a worse prognosis than among the general population [13]. Anal cancer-a human papilloma virus-associated tumor-demonstrated an extremely elevated SMR (ie, 270), an excess that reflects the increased incidence of such illness in people with HIV infection or AIDS [1, 2] and its rarity as cause of death in the general population of the same age of people with AIDS.
 
The spectrum of NADCs that emerged from this analysis was in agreement with those reported by previous studies that took into consideration the frequencies of cancer among the death causes of people with HIV infection [7-9]. In absolute terms, lung cancer was the most common cause of death among the NADCs (58 cases [24.6%]), followed by liver cancer and Hodgkin lymphoma (both with 28 cases [11.9%]), in line with analyses conducted elsewhere [7, 9].
 
Our findings of an excess mortality for NADCs cannot be totally explained by the well-known excess incidence of NADCs among people with AIDS [1, 2, 4, 5]. Some very high SMRs detected in our study should be considered as the joint result of increased incidence of such tumors and of their worse prognosis among people with AIDS versus the general population.
 
We emphasize that our findings refer only to people with AIDS, who may not be representative of HIV-infected people. Because HAART has decreased the incidence of AIDS among HIV-infected individuals, such persons may die of an NADC without developing an AIDS-defining condition. Therefore, our SMRs could be potentially biased with respect to the overall population of HIV-infected people.
 
There are known methodological limitations that arise when comparing causes of death in people with HIV/AIDS against the general population, and several approaches have been employed [6, 8, 14]. For instance, to compute relative risks of cancer-related death in HIV-infected people versus HIV-uninfected people, Selik and Rabkin [6] considered any mention of cancer in death certificates, regardless of whether it was the underlying cause of death. The Cause of Death protocol [8], which classifies the underlying cause of death according to specific rules, could not be used in this analysis, because it would prevent the comparison with the general population of the same sex and age. We acknowledge that ICD-10 rules may be difficult to apply to people with severe immunodeficiency, because these individuals are at risk of dying of a multitude of causes; thus, one may assume that the proportion of people with AIDS classified as having died of NADCs is overestimated. However, it should be noted that about 35% of all NADCs listed in death certificates were not classified, in this study, as the underlying cause of death. The low number of observed and/or expected deaths for certain cancer types was another limitation of this study. Conversely, completeness was the main strength, given the full coverage of the Italian population by the 2 data sources used (ISTAT and National AIDS Registry). Furthermore, for the identification of the underlying cause of death, we adopted the same procedure for both people with AIDS and the general population, thus improving the internal validity of the comparison.
 
In conclusion, this analysis provides estimates of the excess risk of death for NADCs among people with AIDS in the HAART era. Our findings also highlight the importance of monitoring the cancer burden on mortality of people with AIDS, offering specialists in infectious diseases and oncologists an additional means of evaluation of the death threat of NADCs to people with AIDS.
 
 
 
 
  iconpaperstack view older Articles   Back to Top   www.natap.org