Malignancy-related causes of death in human immunodeficiency virus–infected patients in the era of highly active antiretroviral therapy†
Performed on behalf of the Mortalité 2000 Study Group. Scientific coordination: Geneviève Chêne, Thierry May, Philippe Morlat, Dominique Salmon, Dominique Costagliola, and Eric Jougla. Observers: François Dabis, Jean-François Delfraissy, Catherine Leport, and Patrick Yéni. Corresponding physicians: Laurence Héripret, Sibylle Bévilacqua, Fabrice Bonnet, and Charlotte Lewden. Technical team: Jean Boileau, Mallorie Dellac, Sylvie Dutoit, and Valérie Mazou. Technical support: Marthe-Aline Jutand and Gérard Pavillon. Participant wards: Agen (Y. Imbert), Aix-en-Provence (T. Allégre and M. Marquiant), Ajaccio (J.F. Abino), Albi (P. Barel), Alés (A. Lagier), Amiens (J.L. Schmit, J.P. Denoeux, and J.F. Poulain), Angers (J.M. Chennebault and J. Loison), Annecy (J.P. Bru and J. Gaillat), Arcachon (A. Dupont), Argenteuil (M. Pulik), Arras (J.F. Bervar), Avignon (G. Lepeu), Bagnols-sur-Cèze (C. Guglielminotti), Bar-Le-Duc (P. Evon), Basse-Terre (F. Boulard), Bayonne (F. Bonnal), Bazas (M. Amanieu), Beauvais (J.L. Dutel, Y. Courouble, and K. Ghomari), Belfort (J.P. Faller), Besançon (H. Gil, J.M. Estavoyer, B. Hoen, R. Laurent, D.A. Vuitton, G. Achard, and F. Cocquet), Bobigny (L. Guillevin, M. Bentata, B. Jarousse, and P. Honoré-Berlureau), Bondy (M. Thomas and V. Jeantils), Bordeaux (J. Beylot, P. Morlat, F. Bonnet, M. Dupon, M. Le Bras, J.M. Ragnaud, F. Moreau, B. Portal, E. Renoux, F. Terrier, and Y. Guiguen), Boulogne-Billancourt (T. Hanslik), Boulogne (E. Rouveix and H. Berthé), Bourg-en-Bresse (P. Granier), Brest (M. Garré, M.C. Derrien, F. Klotz, and B. Sassolas), Briançon (P. Brousse), Caen (C. Bazin, P. Letellier, and P. Feret), Cayenne (M. Sobesky, P. Coupié, and V. Walter), Château-du-Loir (J.P. Boinet), Clamart (F. Boué and A.M. Delevalle), Clermont-Ferrand (H. Laurichesse), Clichy (B. Fantin), Colmar (N. Plaisance and J.L. Wiederkehr), Colombes (P. Vinceneux and E. Mortier), Compiègne (P. Veyssier and D. Merrien), Corbeil-Essonnes (A. Devidas), Coulommiers (M. Gatfosse), Créteil (A. Schaeffer, A. Sobel, J.D. Magnier, M. Choustermann, and V. Garrait), Dax (P. Loste), Digne-Les-Bains (P. Granet-Brunello), Dijon (H. Portier), Dunkerque (F. Bonnevie and M. Wetterwald), Epinal (H. Jeanmaire), Fort-de-France (G. Sobesky and A. Cabié), Fréjus (E. Counillon), Garches (C. Perronne and J. Salomon), Grenoble (J.P. Stahl and P. Leclercq), La Roche-sur-Yon (P. Perré and O. Aubry), La Rochelle (I. Courbes and E. Brottier-Mancini), Laval (J.C. Hoel), Le Chesnay (J.P. Bedos, J. Doll, J. Laffay, and A. Greder-Brelan), Le Kremlin-Bicêtre (J.F. Delfraissy, C. Goujard, Y. Quertainmont, and M.T. Rannou), Libourne (P. Legendre), Longjumeau (Y. Le Mercier and B. Mougeon), Lons-le-Saunier (D. Baborier), Lure (Y. Selles), Lyon (J.L. Touraine, J.M. Livrozet, C. Trepo, E. Garcia, V. Guéripel, D. Peyramond, and C. Lalain), Mantes-la-Jolie (F. Trémolières), Marmande (J. Testaud), Marseille (J.P. Delmont, J. Moreau, X. Lemaître, G. Fabre, J.A. Gastaut, J. Soubeyrand, T. Gamby, H. Gallais, and V. Lecomte), Mende (P. Meissonier), Metz (P. Bernard and B. Christian), Monaco (B. Taillan), Montfermeil (M. Consoli, M. Echard, and N. Delas), Montpellier (J. Reynes, M. Brunel, V. Faucherre, C. Tramoni, and C. Merle), Mulhouse (G. Beck-Wirth, P. Henon, and M. Benomar), Nanterre (M. Ruel and K. Chemlal), Nantes (F. Raffi and C. Guerbois), Nevers (J.C. Lebas de la Cour and M. Musat), Nice (J.P. Cassuto, P. Dellamonica, and C. Senesi), Nîmes (J.M. Mauboussin), Niort (J.M. Descamps), Nouméa (F. Lacassin), Noyon (F. Grihon), Oloron-Ste.-Marie (M. Begorre), Orléans (T. Prazuck, C. Mille, P. Arsac, and P. Vilanou), Papeete (G. Soubiran), Paris Alfred-Fournier (F. Lunel-Fabiani and M. Cergely), Bichat (J.L. Vildé, P. Yéni, J.P. Coulaud, B. Régnier, E. Bouvet, M. Auburtin, Y. Bennai, C. Gaudebout, and C. Mandet), Boucicaut (B. Patri and P. Bellaiche), Cochin (D. Sicard, D. Salmon, and L. Héripret), Croix-Saint-Simon (G. Raguin), Fernand-Vidal (F. Questel), Georges-Pompidou (M. Kazatchkine, H. Durand, W. Lowenstein, L. Weiss, D. Batisse, M. Marinier-Roger, and D. Tisné-Dessus), Laennec (J.M. Andrieu, F.C. Hugues, and F. Dendoune), Lariboisière (C. Caulin, J.M. Salord, and N. Bonfanty), Moulin-Joly (M. Bary), Necker (B. Dupont and J.P. Viard), Pitié-Salpétrière (S. Herson, A. Simon-Coutellier, F. Bricaire, P. Bossi, L. El Hajj, V. Zeller, and C. Brançon), Rothschild (W. Rozenbaum and S. Thévenet), Saint-Antoine (P.M. Girard, J.C. Imbert, O. Picard, M.C. Meyohas, J.L. Meynard, N. Desplanques, D. Berriot, and B. Gaujour), Saint-Louis (P. Morel, D. Séréni, C. Lascoux-Combes, J.M. Decazes, J.M. Molina, and D. Ponscarme), Saint-Joseph (J. Gilquin and A. Cros), Tenon (C.Y. Mayaud, E. Bergot, M. Wislez, and C. Zurita), Périgueux (P. Lataste and M. Roques), Perpignan (H. Cros), Pessac (J.L. Pellegrin and S. Tchamgoué), Pointe-à-Pitre (M. Strobel), Poissy (H. Masson), Poitiers (B. Becq-Giraudon and G. Le Moal), Pontoise (O. Danne), Quimper (P. Perfezon), Reims (G. Rémy, C. Rouger, and I. Béguinot), Rennes (C. Michelet, C. Arvieux, and M.C. Delmont), Roanne (G. Chaumentin), Rochefort (M.T. Climas), Rouen (F. Caron, I. Gueit, and F. Lecomte), Saint Brieuc (G. Dien and C. Devaurs), Saint Denis (D. Méchali and M.A. Khuong), Saint Denis de la Réunion (C. Gaud), Saint Etienne (F. Lucht, A. Frésard, P. Cathebras, and V. Ronat), Saint Laurent du Maroni (F. Bissuel), Saint Lo (P. Hazera), Saint Mandé (R. Roué and T. Debord), Saint Michel (M. Bonnefoy), Saint Nazaire (C. Micheau), Saint Omer (H. Monnot), Saint Pierre de la Réunion (P. Poubeau), Saintes (T. Pasdeloup), Sarrebourg (E. Grillat), Saverne (F. Loth), Sète (B. Kitschke), Soissons (D. Line), Strasbourg (J.M. Lang, P. Fraisse, G. Ruellan, and P. Fisher), Suresnes (O. Blétry and D. Zucman), Tarbes (J. Petitou), Thionville (M. Grandidier), Toulon (A. Lafeuillade and J.P. de Jaureguiberry), Toulouse (P. Massip and L. Cuzin), Tourcoing (Y. Mouton and F. Ajana), Tours (P. Choutet and J.M. Besnier), Troyes (E. Libbrecht), Valence (R. Riou), Valenciennes (C. Fontier), Vandoeuvre-les-Nancy (T. May and S. Bévilacqua), Vernon (C. Richard), Villejuif (D. Vittecoq, M. Malet, M.H. Salamagne, and C. Boliot), Villeneuve-Saint-Georges (C. Lafaix and O. Patey), and Villeneuve-sur-Lot (E. Buy).
Abstract
BACKGROUND
Before the introduction of highly active antiretroviral therapy (HAART), malignancies accounted for less than 10% of all deaths among human immunodeficiency virus (HIV)-infected patients. This figure may have increased, and the observed types of malignant disease may have been modified, as a result of decreased occurrence of opportunistic infections, the chronicity of HIV infection, the possible oncogenic role of HIV itself, and the aging of the HIV-infected population.
METHODS
All French hospital wards involved in the management of HIV infection were asked to prospectively document the deaths of HIV-infected patients in the year 2000. Underlying causes of death were defined using a standardized questionnaire.
RESULTS
Of a total of 964 deaths, 269 (28%) were attributable to malignancies. Acquired immunodeficiency virus (AIDS)-related malignancies were the underlying cause of 149 deaths (15%); among these malignancies were non-Hodgkin lymphoma (n = 105 [11%]), noncerebral lymphoma (n = 78 [median CD4 count, 86 × 106 per liter; interquartile range [IQR], 35–231 × 106 per liter), and primary cerebral lymphoma (n = 27 [median CD4 count, 20 × 106 per liter; IQR, 4–109 × 106 per liter). Kaposi sarcoma was associated with 40 deaths (4%), and cervical carcinoma was associated with 5 (0.5%). Non-AIDS-related malignancies were the underlying cause of 120 deaths (13%); these non-AIDS-related malignancies included 103 solid tumors (50 respiratory tumors, 19 hepatocarcinomas, 9 digestive tumors, and 6 anal tumors; median CD4 count, 218 × 106 per liter; IQR, 108–380 × 106 per liter) and 17 hemopathies (12 Hodgkin lymphomas, 4 myeloid leukemias, and 1 myeloma; median CD4 count, 113 × 106 per liter; IQR, 56–286 × 106 per liter). Compared with patients who died of other causes, patients who died of solid tumors were more likely to be male, to smoke, to be older, and to have higher CD4 counts.
CONCLUSIONS
Malignant disease has been a major cause of death among HIV-infected patients in industrialized nations since the introduction of HAART. Whereas lethal hemopathies and Kaposi sarcoma are associated with advanced immunosuppression, lethal solid tumors can occur in patients with controlled HIV infection. Cancer 2004. © 2004 American Cancer Society.
The use of highly active antiretroviral therapy (HAART) in industrialized countries has dramatically decreased mortality rates for human immunodeficiency virus (HIV)-infected patients since 1996 and has led to a shift in causes of death in this population.1-3 Cohort studies have shown that concomitant with the dramatic decrease in AIDS-related causes of death among HIV-infected patients, there has been an increase in the proportion of non-AIDS-related causes of death, such as hepatitis B and C, cardiovascular diseases, iatrogenic disorders, and (primarily) malignant disease.4-6 Before HAART was in widespread use among patients with HIV, non-AIDS-related malignancies accounted for less than 1% of all causes of death in this population7-9; the subsequent dramatic increase in this figure may be attributable to the modification of competitive risks, the chronicity of HIV infection, the possible oncogenic role of hepatitis B and C viruses as well as HIV itself, and the aging of the HIV-infected population. Furthermore, AIDS-related causes of death may also have been affected by the introduction of HAART. For example, the incidence of non-Hodgkin lymphoma (NHL) has decreased more slowly than has the incidence of various AIDS-related opportunistic infections, and as a result, NHL may be a major cause of morbidity and mortality in the HAART era.2, 4 To date, no large-scale study of the distribution of lethal malignancies in the HAART era has been performed, and thus the extent to which the management of HIV-infected patients could be suitably modified remains unknown. The current study focuses on malignancy-related death in HIV-infected patients and describes the distribution and characteristics of these lethal malignancies using data obtained in a national survey conducted in 2000 in France.
MATERIALS AND METHODS
‘Mortalité 2000’ was a national survey investigating the causes of death among HIV-infected patients in France in the year 2000. All hospital wards known to be involved in the case management of HIV infection in France were contacted. A mailing list was compiled from several sources involved in organizing the management of HIV-infected patients in France. Participants were contacted at the beginning of each trimester in 2000 and were asked to prospectively document deaths and supposed causes of death for HIV-infected adults (age ≥ 18 years). Deaths were documented using a standardized questionnaire that examined sociodemographic characteristics (gender, date and place of birth, date and place of death, and socioeconomic status), background and comorbidities (alcohol consumption, smoking, drug use, serologic status with respect to hepatitis B and C, cirrhosis, hepatocarcinoma, and pathologic type of malignancy), and variables related to HIV infection (date of diagnosis of HIV infection, transmission group, Centers for Disease Control [CDC] clinical stage, latest plasma HIV RNA load, latest CD4-positive T lymphocyte (CD4) count, and previous antiretroviral treatment). The underlying cause of death was defined as the disease or injury that initiated the sequence of morbid events leading to death; this definition was consistent with the Tenth Revision of the International Classification of Diseases (ICD-10).10 One underlying cause of death was selected for each patient and was coded according to the ICD-10 at the French referral center for coding of death certificates (INSERM-CépiDc, Le Vésinet, France).
Malignancies were classified as being AIDS related when active pathologies at death included one AIDS-defining malignancy according to the 1993 CDC clinical classification for HIV infection.11 The following were considered to be AIDS-defining malignancies: high-grade NHL, including Burkitt and immunoblastic lymphoma; primary brain lymphoma; Kaposi sarcoma; and cervical carcinoma. Other types of malignant disease, as well as low-grade NHL, were classified as being non-AIDS-related events.
The characteristics of patients who died of NHL were compared with those of patients who died of other causes and then with the characteristics of patients who died of other AIDS-related causes. For this analysis, we selected only patients with systemic NHL and excluded patients with primary brain lymphoma, because of the different pathogenic process associated with the latter malignancy and because of its association with very low CD4 cell counts (< 50 × 106 per liter) and poor median survival (4 months).12
Solid tumors, excluding hemopathies, were defined as non-AIDS-related malignancies. We compared patients who died of solid tumors with those who died of non-malignancy-related causes to identify comorbidities associated with fatal malignancies.
The chi-square test was used for comparisons involving categoric variables, and the Kruskal–Wallis test was used for comparisons involving continuous variables. SAS software (Version 8.2; SAS Institute, Cary, NC) was used to perform the statistical analyses. Authorization for these analyses was obtained from the French National Commission on Informatic Freedom and Rights.
RESULTS
Overall, 185 of the 195 wards contacted participated in the study; these wards accounted for approximately 64,000 HIV-infected patients who received follow-up in 2000. Participating wards reported 964 deaths among HIV-infected patients in 2000. A standardized questionnaire was completed for 924 of these deaths (96%). Median patient age at death was 41 years (interquartile range [IQR], 36–49 years), and 78% of patients who died were men. Among patients who died, HIV transmission occurred via heterosexual contact in 34% of cases, intravenous drug use in 28% of cases, and homosexual contact in 27% of cases. The median time to death from diagnosis of HIV infection was 7.6 years (IQR, 3.3–11.6 years), and the median CD4 cell count was 94 per microliter (IQR, 19–260 per microliter).
The distribution of primary underlying causes of death is summarized in Table 1, along with the distributions reported in previous studies.3-5, 13 Overall, malignancy-related causes, including AIDS-related (n = 149) and non-AIDS-related causes (n = 120), accounted for a total of 269 deaths (28%) (Table 2). Patients who died of malignant disease had a median age of 43 years at death (IQR, 38–51 years) and a median known duration of HIV infection of 8.0 years (IQR, 3.7–11.8 years). Eighty-three percent of these patients were male, and HIV transmission occurred via homosexual contact in 37% of cases, heterosexual contact in 32% of cases, intravenous drug use in 19% of cases, and other transmission methods in 12% of cases. The median CD4 count at the last available measurement was 110 × 106 per liter (IQR, 34–262 per liter), and the median HIV RNA load (in log[copies/mL]) at last measurement was 3.3 (IQR, 2.2–5.0). Twenty percent of patients had both a CD4 count > 200 × 106 per liter and an HIV RNA load < 500 copies/mL. Positive serologic findings for hepatitis C virus were reported in 24% of these patients, and positive findings for hepatitis B surface antigen were reported in 14%. Ninety percent of patients who died of malignant disease had been treated with antiretroviral therapy (median duration, 4.5 years; IQR, 2.1–7.3 years).
| Mortalité 2000 (Lewden et al., 20003) | Bonnet et al., 20024 | Rosenthal et al., 200313 | Mocroft et al., 20025 | |
|---|---|---|---|---|
| Year of death | 2000 | 1998–1999 | 2001 | 2000–2001 |
| No. of deaths | 964 | 107 | 265 | 413 |
| Underlying cause of death (%) | ||||
| AIDS (NHL) | 47 (11) | 49 (14) | 49 (NA) | 17 (NA) |
| Non-AIDS-related malignancy | 12 | 12 | NA | NA |
| Hepatitis B/C-related terminal-stage liver disease | 9 | 14 | 14 | 9 |
| Cardiovascular disease | 7 | 10 | NA | 6 |
| Bacterial infection | 6 | 3 | NA | NA |
| Suicide | 4 | 5 | NA | 2 |
| Other | 11 | 5 | 37 | 37 |
| Unknown | 3 | 2 | NA | 29 |
- HIV: human immunodeficiency virus; HAART: highly active antiretroviral therapy; AIDS: acquired immunodeficiency syndrome; NHL: non-Hodgkin lymphoma; NA: not assessed.
| Malignancy | No. of cases (% of all malignancies) | % occurring in men | Median CD4 T lymphocyte count (interquartile range) in cells per microliter |
|---|---|---|---|
| Total | 269 (100) | 83 | 110 (34–262) |
| AIDS-related malignancies | 149 (55) | 81 | 60 (15–145) |
| Non-Hodgkin lymphoma | 78 (29) | 84 | 86 (35–231) |
| Primary brain lymphoma | 27 (10) | 65 | 20 (4–109) |
| Kaposi sarcoma | 40 (15) | 95 | 22 (4–64) |
| Cervical neoplasia | 4 (1) | 0 | 447 (261–606) |
| Non-AIDS-related malignancies | 120 (45) | 86 | 186 (97–356) |
| Solid tumors | 103 (38) | 90 | 218 (108–380) |
| Respiratory malignancya | 50 (19) | 88 | 262 (132–443) |
| Hepatocarcinoma | 19 (7) | 83 | 157 (110–353) |
| Digestive malignancyb | 9 (3) | 100 | 175 (24–323) |
| Anal malignancy | 6 (2) | 100 | 137 (118–196) |
| CNS malignancy | 4 (1) | 100 | 121 (108–405) |
| Other | 15 (6) | 67 | 234 (89–380) |
| Hemopathies | 17 (6) | 82 | 113 (56–286) |
| Hodgkin lymphoma | 12 (4) | 90 | 86 (53–147) |
| Otherc | 5 (2) | 80 | 411 (221–541) |
- HIV: human immunodeficiency virus; AIDS: acquired immunodeficiency syndrome; CNS: central nervous system.
- a Lung (n = 41), ear/nose/throat (n = 6), or pleural (n = 3).
- b Pancreatic (n = 3), colonic (n = 3), gastric (n = 1), esophageal (n = 1), or cholangiocarcinoma (n = 1).
- c Myeloid leukemia (n = 4) or multiple myeloma (n = 1).
AIDS-related malignancies accounted for 16% of all deaths. Among these malignancies, NHL was the leading cause of death (n = 105 [11% of all deaths and 23% of all AIDS-related deaths], including 27 cases of primary brain lymphoma) (Tables 2, 3). Patients who died of primary brain lymphoma had a lower median CD4-positive T lymphocyte count than did those who died of NHL (20 × 106 per liter [IQR, 4–109 × 106 per liter] vs. 86 × 106 per liter [IQR, 35–231 × 106 per liter]; P < 0.05). Pathologic types of NHL included diffuse large B-cell lymphoma (23%), Burkitt lymphoma (19%), immunoblastic lymphoma (5%), other pathologic types (13%), and undocumented type (40%). Kaposi sarcoma was associated with 40 deaths (4%), and cervical neoplasia was associated with 5 (0.5%).
| Characteristic | Cause of death | |||
|---|---|---|---|---|
| NHL (n = 78) | PBL (n = 27) | Hodgkin lymphoma (n = 12) | Other hemopathya (n = 5) | |
| Qualitative variablesb | ||||
| Male gender | 84 | 65 | 91 | 80 |
| HIV transmission method | ||||
| Heterosexual contact | 35 | 50 | 18 | 20 |
| Homosexual/bisexual contact | 35 | 23 | 46 | 80 |
| Intravenous drug use | 17 | 12 | 27 | 0 |
| Other/undetermined | 13 | 15 | 9 | 0 |
| Poor socioeconomic statusc | 15 | 17 | 10 | 0 |
| HCV infection | 23 | 15 | 27 | 0 |
| HBs antigen positivity | 7 | 0 | 36 | 0 |
| No previous antiretroviral therapy | 13 | 12 | 0 | 20 |
| No HAART | 16 | 12 | 0 | 20 |
| Quantitative variablesd | ||||
| Age (yrs) | 42 (37–48) | 42 (37–45) | 37 (35–40) | 49 (47–54) |
| Duration of HIV infection (yrs) | 7.0 (1.8–11.6) | 7.3 (2.5–9.4) | 8.3 (3.3–13.4) | 9.0 (8.6–13.2) |
| CD4-positive lymphocyte count (cells/μL) | 86 (35–231) | 20 (4–109) | 86 (53–147) | 411 (221–541) |
| HIV RNA load (log[copies/mL]) | 3.6 (2.3–5.2) | 5.1 (2.5–5.6) | 2.3 (1.7–2.8) | 1.6 (1.4–2.8) |
- NHL: non-Hodgkin lymphoma; PBL: primary brain lymphoma; HIV: human immunodeficiency virus; HCV: hepatitis C virus; HBs: hepatitis B surface; HAART: highly active antiretroviral therapy.
- a Myeloid leukemia (n = 4) or multiple myeloma (n = 1).
- b Values presented are percentages of patients within each cause-of-death group.
- c Characterized by at least one of the following: lack of health insurance, lack of employment, lack of housing accommodations, income < 535 euros per month, and illegal immigrant status.
- d Values presented are median values with interquartile ranges in parentheses.
Patients who died of NHL had a higher median CD4 T lymphocyte count than did patients who died of other AIDS-related causes (86 × 106 per liter [IQR, 35–231 × 106 per liter] vs. 20 × 106 per liter [IQR, 6–67 × 106 per liter]; P < 10−4), and the former group also had a lower HIV RNA load (median log[copies/mL], 3.6 [IQR, 2.3–5.2] vs. 5.1 [IQR, 4.2–5.6]; P < 10−4). Compared with all other patients who died, patients who died of NHL were less likely to have had poor socioeconomic status (15% vs. 35%; P < 10−4) and to have been coinfected with hepatitis C virus (24% vs. 36%; P = 0.03). Patients who died of NHL did not differ from other patients in terms of age, gender, HIV transmission method, known duration of HIV infection, median CD4 T lymphocyte count, HIV RNA load, or treatment with HAART (data not shown).
Non-AIDS-related malignancies were associated with 120 deaths (26% of non-AIDS-related deaths and 12% of all deaths) (Table 2). The median CD4 count for patients who died of non-AIDS-related malignancies was 186 × 106 per liter (IQR, 97–356 × 106 per liter). Solid tumors, including 19 cases of hepatitis-related hepatocarcinoma (associated with hepatitis C [n = 8], hepatitis B [n = 7], hepatitis B and C [n = 2], or alcohol use [n = 2]), accounted for 103 deaths, and hemopathies (Hodgkin lymphoma in 12 cases, myeloid leukemia in 4 cases, and multiple myeloma in 1 case) accounted for 17 deaths. Solid tumors identified as the cause of death in patients with HIV infection also were commonly found in respiratory sites (n = 50, including 41 lung tumors, 6 eye/nose/throat [ENT] tumors, and 3 pleural tumors). The pathologic type distribution for the 41 lung tumors was as follows: adenocarcinoma (n = 13), epidermoid (n = 9), anaplastic (n = 5), other (n = 4), and undocumented (n = 10). ENT malignancies included one adenocarcinoma, one epidermoid malignancy, and three carcinomas. Fatal solid tumors also were found in digestive sites (n = 9, including 3 pancreatic adenocarcinomas, 3 colon adenocarcinomas, 1 gastric adenocarcinoma, 1 cholangiocarcinoma, and 1 esophageal carcinoma), the rectum (n = 6, including 5 epidermoid malignancies and 1 neuroendocrine carcinoma), and the central nervous system (n = 4, including 2 glioblastomas). Other sites of fatal solid tumors included prostate (n = 3), breast (n = 3), skin (n = 2), uterus (n = 1), bladder (n = 1), penis (n = 1), muscle (n = 1), and undocumented sites (n = 3).
Table 4 summarizes the characteristics of patients who died of solid tumors, other malignancies (AIDS-related and non-AIDS-related), and other causes. Patients who died of solid tumors were more likely to be male, to smoke, to be older, to have a longer known duration of HIV infection, and to have a higher CD4-positive T lymphocyte count compared with patients who died of non-malignancy-related causes.
| Characteristic | Cause of death | |||
|---|---|---|---|---|
| Solid tumora (n = 103) | Other malignancyb (n = 166) | Other causes (n = 695) | Pc | |
| Qualitative variablesd | ||||
| Male gender | 86 | 82 | 76 | 0.03 |
| HIV transmission method | 0.92 | |||
| Heterosexual contact | 36 | 29 | 35 | |
| Homosexual/bisexual contact | 25 | 44 | 23 | |
| Intravenous drug use | 28 | 14 | 31 | |
| Other/undetermined | 10 | 12 | 11 | |
| Poor socioeconomic status | 25 | 16 | 38 | 0.01 |
| Tobacco use | 71 | 38 | 53 | 10−3 |
| Alcohol use | 35 | 9 | 32 | 0.66 |
| HCV infection | 31 | 19 | 39 | 0.14 |
| HBs antigen | 19 | 10 | 11 | 0.04 |
| No previous antiretroviral therapy | 7 | 12 | 16 | 0.02 |
| No HAART | 18 | 13 | 24 | 0.18 |
| Quantitative variablese | ||||
| Age (yrs) | 46 (40–56) | 41 (36–47) | 40 (36–48) | < 10−4 |
| Duration of HIV infection (yrs) | 9.0 (5.8–12.6) | 7.5 (1.8–11.0) | 7.4 (3.2–11.5) | 0.02 |
| CD4-positive T lymphocyte count (cells/μL) | 218 (108–380) | 64 (15–155) | 82 (15–258) | < 10−3 |
| HIV RNA load (log[copies/mL]) | 2.6 (1.7–4.2) | 3.9 (2.3–5.4) | 4.5 (2.3–5.4) | < 10−3 |
- HIV: human immunodeficiency virus; HCV: hepatitis C virus; HBs: hepatitis B surface; HAART: highly active antiretroviral therapy.
- a Includes all non-AIDS-related malignancies except for hemopathies.
- b Includes all AIDS-related malignancies (non-Hodgkin lymphoma, primary brain lymphoma, Kaposi sarcoma, and cervical neoplasia) and all hemopathies (Hodgkin lymphoma, myeloid leukemia, and myeloma).
- c For comparison of the Solid tumor group with the Other causes group.
- d Values presented are percentages of patients within each cause-of-death group.
- e Values presented are median values with interquartile ranges in parentheses.
DISCUSSION
The current prospective study, which was designed specifically to collect data on causes of death among HIV-infected adults over the course of a year during the HAART era, found that malignant disease was the most common cause of death. Overall, malignant disease (AIDS-related or non-AIDS-related) accounted for 28% of deaths in the study cohort. This frequency remains lower than the corresponding frequency in the general population of France (38% for persons ages 20–59 years in 1999),14 and the distribution of malignancy types also differs between the study cohort and the general population. Malignancy-related deaths were relatively common among HIV-infected patients who were mildly immunocompromised; 26% of patients who died of NHL and 53% of patients who died of solid tumors had CD4 T lymphocyte counts > 200 × 106 per liter.
The incidence of NHL has decreased dramatically since the introduction of HAART,2, 15, 16 but deaths related to other immunoproliferative malignancies continue to be more common among HIV-infected patients compared with the overall French population (11.5% vs. < 3%).14 Using a European database on HIV-infected patients, Mocroft et al.2 demonstrated that the incidence of AIDS-defining events decreased from 30.7 per 100 patient-years in 1994 to 2.5 per 100 patient-years in 1998, whereas the proportion of AIDS diagnoses that were attributable to NHL increased over that period. Among the different types of NHL, the decrease in incidence has been more pronounced for primary brain lymphoma than for systemic AIDS-related NHL. Besson et al.15 reported a decrease in the incidence of primary brain lymphoma from 27.8 per 10,000 person-years to 9.7 per 10,000 person-years between 1993–1994 and 1997–1998, compared with a decrease in the incidence of systemic NHL from 86 per 10,000 person-years to 42 per 10,000 person-years. The incidence of systemic NHL within each stratum of patients with similar CD4 counts did not decrease between 1993–1994 and 1997–1998, suggesting that a decrease in the proportion of patients with low CD4 counts may have been responsible for the observed change in systemic NHL incidence. Another possible explanation for the relatively high frequency of NHL-associated death is that despite improvements in survival since the introduction of HAART,15, 17 the prognosis for patients with systemic NHL remains poor, with a median survival duration of 6 months and a 2-year survival rate of 41%.18-20 In this new epidemiologic context, the determination of risk factors and the identification of early markers of NHL appear to be important. The known risk factors for AIDS-related NHL in the HAART era are age, prolonged HIV infection, and prolonged immunodeficiency and B-cell stimulation (as indicated by increased serum globulin levels).21, 22 The roles of HIV and certain coinfecting viruses (especially Epstein–Barr virus and hepatitis C virus) require further assessment in large, prospective immunohistopathologic studies.
Fatal Hodgkin lymphoma also is overrepresented among the HIV-infected population (accounting for 1.2% of deaths in the current study vs. 0.4% of deaths among persons ages 20–59 years in the general population in 1999), although the prognosis for patients with Hodgkin lymphoma, which is not considered an AIDS-related event, has improved significantly since the introduction of HAART.23 It is noteworthy that the median CD4-positive T lymphocyte count for patients who died of Hodgkin lymphoma was similar to the corresponding figure for patients who died of NHL.
The proportion of solid tumors considered to be the cause of death has increased steadily over time among HIV-infected patients. Before the widespread use of HAART, such tumors were found to account for < 1% of all deaths in cohort studies,2, 7, 24 compared with > 10% of all deaths in the current study. Nonetheless, this change appears to be related primarily to the decrease in the incidence of AIDS-defining malignancies since the introduction of HAART and not to an increase in the incidence of other malignancies.25 Given the poor prognosis for HIV-infected patients with malignant disease (particularly lung carcinoma26), the high level of exposure to risk factors (e.g., tobacco use, alcohol use, viral coinfection, inadequate nutrition, and immunodeficiency) in this population, and the fact that patients in this population are aging, solid tumors may become a major cause of morbidity and mortality among HIV-infected individuals in the coming years.
The issue of lung carcinoma demands particular emphasis, due to its poor prognosis and due to the relatively high percentage of lung carcinoma–related deaths in the current study (5% of all deaths and 15% of malignancy-related deaths). This finding may be attributable to the highly aggressive nature of lung carcinoma, to late diagnosis in some patients, and to suboptimal treatment,26, 27 or it may simply be attributable to increased incidence.28, 29 Nonetheless, the rate of lung carcinoma–related death in the current cohort remains slightly lower than the corresponding rate in the general population of France (22.5% of all malignancy-related deaths in 1999), suggesting that the proportion of deaths related to lung carcinoma may increase in the near future, given the expected decrease in immunoproliferative malignancies and the increased life expectancy of HIV-infected patients. Thus, controlling exposure to risk factors (particularly tobacco use) is one of the main challenges in preventing HIV-infected patients from developing solid malignancies.
In the current study, 53% of patients who died of solid tumors and 53% of patients who died of hepatitis-related hepatocarcinoma had CD4 T lymphocyte counts > 200 × 106 per liter. We do not have data regarding the specific treatment administered to these patients, but we speculate that they did not receive optimal treatment for their HIV seropositivity, a finding that has been reported in other case series.26 The prognosis for HIV-infected individuals can reasonably be expected to improve in the near future, given the general increase in CD4 T lymphocyte counts in the HIV-infected population and the advancements made in the specific management of patients with HIV.
It remains unknown as to whether HIV itself has an independent effect on carcinogenesis. Some epidemiologic studies report an increased risk of lung carcinoma in HIV-infected patients, but data on tobacco use, the major confounding factor in such studies, are always sparse.30-33 Some in vitro studies have shown that the HIV tat gene may modulate the expression of certain protooncogenes (c-myc, c-fos, and p53) in malignant cells.34 The role of HIV itself in the development of solid malignancies, if it exists, probably is relatively minor compared with the impact of risk factors such as tobacco use, alcohol use, poor nutrition, coinfection with other protooncongene viruses (e.g., hepatitis, human papillomavirus, etc.), and immunodepression, which often is present in HIV-infected patients and should be the target of preventive and curative strategies.
In clinical practice, systematic screening for tobacco and alcohol abuse should be implemented, along with screening for lung carcinoma (via annual X-ray examination) and for cervical and rectal malignancies (via systematic examination, cervical smears, and human papillomavirus detection); in addition, the development of early markers of NHL is a priority. In addition, the management of malignancies in HIV-infected patients should be optimized in an effort to improve prognosis.
In conclusion, malignant disease appears to be a major cause of death among HIV-infected patients in industrialized nations during the HAART era. Whereas most lethal cases of hemopathy and Kaposi sarcoma are associated with advanced immunosuppression, solid tumors can affect patients with controlled HIV infection. The incidence of fatal solid tumors may grow in the coming years, due to increases in life expectancy and the duration of HIV seropositivity for HIV-infected patients treated with HAART.
Acknowledgements
The authors thank Ray Cooke for his very helpful revision of the current article.
