- Email: [email protected]
HIV, occupational exposure, and medical responsibilities
clinical and applied medicine to be published in your journal. I am, of course, encouraged by Horton’s statement (July 1, p 3) that "the unifying science of medicine is an inclusive science of interpretation, one that 7he Lancet will nourish with enthusiasm". My plea is that The Lancet, in assessing its own pertinent role in modern scientific journalism should continue to assert positively its heritage in clinical and applied medicine.
Anthony G Freeman Meadow Rise, 3 Lakeside, Swindon, Wiltshire SN3
1QE, UK
Can antiviral agents decrease the of Kaposi’s sarcoma?
*Considered as time-dependent covariates. Table: Relative hazards (95% CI) and p value for KS in Cox proportional hazards model of time to KS
occurrence
SiR-Following the publication by Chang et all of the identification of herpesvirus-like DNA sequences in AIDSassociated Kaposi’s sarcoma (KS), several teams have reported similar results in The Lancet (March 18, p 722-23; March 25, p 759-61 and p 761-62; April 22, p 1043 and p 1043-44; May 6, p 1180) and elsewhere.2 We studied whether antiviral agents active against herpesvirus could reduce the occurrence of KS, as postulated by Jones et aP and in line with the report by Morfeldt and Torsander.’ The French clinical epidemiology database on HIVseropositive subjects followed in hospitals was started in 1989 in more than 40 hospitals in France. The standardised data collection form includes characterisation of exposure of usual biological markers, clinical group, values manifestations of HIV infection (physician’s diagnosis), nature of treatments delivered, identification of clinical trials in which the patient was included, and death and cause of death as reported in the medical records. A follow-up form is used at each visit or hospital admission at which a clinical manifestation of HIV infection is reported or a new treatment is prescribed, or at least every 6 months. Entry in this study was defined as the time of the first recorded CD4 count. Only patients with a follow-up equal to or longer than 6 months were included. Exclusion criteria were KS at entry or previous reported history of KS and participation in a trial evaluating antiretroviral treatment and/or acyclovir, foscarnet, or ganciclovir. Among the 20 963 subjects included in the database, 16 229 fulfilled the study criteria, 76% were men, and 41 were men who have sex with men. The median follow-up time, from entry to KS or last visit, 20 months (interquartile range 13-33). KS was was diagnosed in 1106 patients at some time during follow-up (7%), and for 48% KS was the first AIDS-defining illness. More than two-thirds of patients (69%) were prescribed antiretroviral treatment before the end of the study period; (16%) were prescribed acyclovir, 782 (5%) ganciclovir, (0-8%) foscarnet. Data were analysed with Cox proportional hazards regression model with time-dependent 2626
and 135
covariates. Results are presented in the table. The relative hazard for KS was slightly increased with acyclovir (1-3 in our study vs 1-4 in Jones et aP); no difference was found with ganciclovir (relative hazard was 1-0 in both studies) or foscarnet (relative hazards 1-4 in our study compared with 0-3 in Jones et al). As in the report by Jones and colleagues, results were similar when the analysis was restricted to men, or to men who have sex with men. Therefore, the results of both studies, which had similar design and analysis, were similar, although the follow-up was longer in our study and fewer medications were prescribed, with the exception of foscarnet for which the two studies are discordant. It should be kept in mind that both studies were observational and therefore potentially subject to 578
comparison bias. In our view, one such bias could explain slight increase in the risk of developing KS after acyclovir. It has been reported that patients who take both zidovudine and acyclovir survive longer. They are therefore more likely to develop disease as their time at risk increases. Could another bias acting in an opposite direction in each study explain the difference? Other evaluations of the relation between Kaposi’s sarcoma and antiviral medications active against herpesviruses are needed to clarify the issue.
the
*D Costagliola, M Mary-Krause, for Clinical Epidemiology Group from Centres d’Information et de Soins de l’Immunodéficience Humaine B3E, INSERM SC4, Institut Fédératif Saint-Antoine de Recherches en Santé, Faculté de Médecine Saint-Antoine, Université Pierre et Marie Curie, 75571 Paris, France
1
2
3 4
5
Chang Y, Cesarman E, Pessin MS, et al. Identification of new human herpesvirus-like DNA sequences in AIDS-associated Kaposi’s sarcoma. Science 1994; 266: 1865-69. Ambroziak JA, Blackbourn DJ, Herndier BG, et al. Herpes-like sequences in HIV-infected and uninfected Kaposi’s sarcoma patients. Science 1995; 268: 582-83. Jones JL, Hanson DL, Chu SY, Ward JW, Jaffe HW. AIDS-associated Kaposi’s sarcoma. Science 1995; 267: 1078-79. Morfeldt L, Torsander J. Long-term remission of Kaposi’s sarcoma following foscarnet treatment in HIV-infected patients. Scand J Infect Dis 1994; 26: 749-52. Cooper DA, Pehrson PO, Pedersen C, et al. The efficacy and safety of zidovudine alone or as cotherapy with acyclovir for the treatment of patients with AIDS and AIDS-related complex: a double-blind randomized trial. AIDS 1993; 7: 197-208.
HIV, occupational exposure, and medical responsibilities management of clinically significant HIV infection requires knowledge of the exposure maximum period between exposure and the development of detectable antibodies in serum. Meyohas and colleagues (June 24, p 1634) suggest that this time could be greater than 6 months, and that current practice should be changed. Before such changes are introduced, the criteria used to define the duration of the seroconversion interval should first be established. These criteria, at the least, should include documentation of (a) a contemporaneously recorded significant exposure to (b) an identifiable source, with the source having (c) laboratory-confirmed evidence of HIV, and where the recipient is (d) shown to be uninfected at the time of the exposure, and (e) shown, through interval testing instigated in response to the reported exposure, to have become infected. The test used to establish the time of the last negative HIV test should be equivalent to that used to determine the time of the first positive test. Reasonable effort should be made to exclude a second, later exposure, including, where relevant, documentation of the HIV status
SiR-Appropriate to
Figure: Periods in which HIV seroconversion took place after occupational exposure 37 percutaneous and four mucocutaneous exposures. Previous
ot sexuai contacts. wnere
cases:
12 in USA, 29 elsewhere.
possiole, Miv 1rom tne putative and the infected recipient should be compared by molecular techniques. These criteria are most likely to be fulfilled after occupational exposure, and review of the 73 cases of definite occupationally acquired HIV infection reported worldwide’I is instructive. In 41 patients, the time of the first positive result of a test for HIV antibody was reported (figure); only in three was anti-HIV first demonstrated at more than 6 months after exposure. One was detected at blood donation 9 months after mucocutaneous exposure, having tested negative at 8 weeks post exposure; a second exposure, to a phlebotomy needle used on an injecting drug user of unknown HIV status, happened 2 months after the first.2 The second case was detected when contacted for testing with a new anti-HIV test in 1986, 11 months after percutaneous exposure to blood from an injecting drug user who was anti-HIV negative but had detectable p24 antigen.3 In the third, seroconversion took place between day 90 and day 288.4 None of the three recalled a seroconversion illness. One had tested anti-HIV negative at 3 months post exposure, but none had been tested at 6 months. 29 (63%) of 46 cases described in detail had symptoms compatible with acute HIV infection. Most symptoms (in 26, 90%) had arisen between the 2nd and 6th weeks after exposure. One illness occurred more than 3 months later: after exposure in November, 1985, the worker tested anti-HIV negative for 3 months, developed symptoms in April, 1986, and was proven infected in May, 1986. A second exposure was not excluded.5 Meyohas and colleagues do not report the infection source in their patient, and the combination of late onset of possible symptoms and late seroconversion is unusual (figure). Further details about the exposure, the source, and the process by which a later exposure was excluded would help in assessment of the public health importance of this event. The transmission rate after a single percutaneous exposure to a known HIV infected source has been estimated as being around 1 in 300: most of those so exposed will not become infected. Achieving compliance with anti-HIV testing even at baseline, 6 weeks, 3 months, and 6 months post exposure is difficult. We believe that there is insufficient evidence to justify change in what has inevitably become, despite source
guidance recommending testing exposure, common practice. *Julia
at
12
months
after
Heptonstall, O Noel Gill
PHLS Communicable Disease Surveillance Centre, London NW9 5EQ, UK
1
Heptonstall J, Porter KP, Gill ON. Occupational transmission of HIV: summary of published reports to July 1995. London: Public Health 1995 (internal report). Centers for Disease Control. Update: human immunodeficiency virus infections in health care workers exposed to blood of infected patients. MMWR 1987; 36: 285-89. Ippolito G, Puro V, de Carli G, and the Italian Study Group on Occupational Risk of HIV Infection. The risk of occupational human immunodeficiency virus infection in health care workers. Arch Int Med 1993; 153: 1451-58. Ramsey KM, Smith EN, Reinarz JA. Prospective evaluation of 44 health care workers exposed to human immunodeficiency virus-1, with one seroconversion. Clin Res 1988; 36: 22A. Wallace MR, Harrison WO. HIV seroconversion with progressive disease in health care worker after needlestick injury. Lancet 1988; i: 1454.
Laboratory Service, 2
3
4
5
and colleagues’ report of HIV seroconversion after needlestick injury in a woman deserves comment. Because medical information on HIV and AIDS is largely disseminated by the media, we should be prudent before publishing such clinical cases. No rigorous scientific conclusion can be drawn from Meyohas’ report because the presumed absence of risk factors for HIV, other than needlestick injury, between May and November, 1993, is based solely on the reliability of questioning. That the needle was used in an HIV-infected patient is not certain, and thus no information is available on the patient’s clinical and biological status. Comparison of nucleotide sequence between HIV isolates from the patient and the woman would have provided true evidence for a connection between the needlestick injury and the syndrome of primary HIV infection, as it has been in other circumstances.’ Without these data it is very hazardous to drawn public health conclusions from this observation, as Meyohas and colleagues have done. In France, this case report was publicised 2 days before we received the issue of The Lancet concerned. This coverage caused much anxiety in nurses who had been accidentally exposed to HIV in past months. Finally, we know that the time between submission of and
SIR-Meyohas
579
letter is short, so we do not understand why this case was not reported at the end of 1993, especially since it is of such public health importance.
publication of a
*Alain Lafeuillade, Catherine Thiebaut, Robert Quilichini, Nérina Profizi *Department of Infectious Diseases and Laboratory of Virology General Hospital, 83056 Toulon, France
1
Jaffe HW, McCurdy JM, Kalish ML,
et al. Lack of HIV transmission in the practice of a dentist with AIDS. Ann Intern Med 1994; 121: 855-59.
authors’
reply
SiR-In
Heptonstall’s report, a small number of patients negative for anti-HIV antibody beyond the third month after exposure to HIV. As in our findings, the existence of a delayed preseroconversion period is difficult to establish formally. However, it cannot be excluded in rare cases. Our only aim, when publishing our findings in The Lancet, was to ask of the scientific community whether other similar cases had been observed worldwide. We reported this observation only for the sake of vigilance in public health. Lafeuillade raises the issue of worrying the general population with such publication. The cases of HIV-0 infection, reported in The Lancet several months ago, could also worry the general population in so far as they might be an underestimate of the diagnosis of infection. However, the publication of our findings was useful. Our observation appeared sufficiently troubling to report to the scientific community. We chose to wait until 1995 before publishing our findings, to ensure through long follow-up and continued interviews that the patient did not have other risk factors of HIV infection during the critical period.
were
M C
Meyohas,
L
Morand-Joubert, *J J Lefrère
INTS Hôpital St-Antoine, 75012 Paris, France
Table: Multivariate
Chemotherapy for elderly patients with lung cancer
SiR-More than 50% of patients with inoperable non-small cell lung cancer (NSCLC) are over the age of 65, yet palliative chemotherapy, if given at all, is usually reserved for younger age groups. Are there any sound data for denying these elderly patients treatment? The upper age limit in only one of eight authoritative randomised trials of chemotherapy versus supportive care for patients with advanced NSCLC extended to 75 years, and the average of the median ages in this group of trials was about 60 years (reviewed in ref 1). There was no evidence from these trials that the elderly fared less well with chemotherapy. The impact of age as a prognostic factor for survival and response in advanced NSCLC has been analysed in reports totalling 4516 patients2.3 (and reviewed in ref 3). In one of these, the SouthWest Oncology Group analysed a database of 2531 patients with stage IV NSCLC2 and found by multivariate analysis that age greater than 70 years was in fact a significant independent positive predictor for survival (p=002). In the remainder of these studies age had no influence by multivariate analysis. Since 1990 we have prospectively generated a database of patients with a proven diagnosis of inoperable NSCLC, referred to the lung unit at the Royal Marsden Hospital. We have now analysed 39 pretreatment variables for their influence on objective response, symptomatic response, and survival to chemotherapy (usually cisplatin-based) in 290 patients treated in a series of phase II trials. The table lists 580
factors identified
analysis for influential prognostic factors
by multivariate analysis
with the stepwise method. Consistent with other studies/,3
logistic regression performance status (Eastern Cooperative Oncology Group),
disease extent, and certain biochemical features were influential variables for survival and objective response, and we also show that this holds for symptom relief-a crucial outcome for treatment given with palliative intent. Additionally, older age was positively associated with objective response. Age had no influence, either positive or negative, on symptom relief or survival. These data indicate that older patients fare just as well as younger ones with palliative chemotherapy, and possibly better. Given the age distribution of NSCLC and the growing elderly population, palliative treatment of the elderly with this disease will increasingly be an issue. We consider that elderly patients with advanced NSCLC should no longer be excluded from palliative chemotherapy trials. T F Hickish, *I E Smith, S Ashley, G Middleton Lung Unit, Royal Marsden NHS Trust, Sutton, Surrey SM2 5PT, UK
1
Future directions in the treatment of non-small cell lung Semin Oncol 1994; 21: 48-62. Albain KS, Crowley JJ, LeBlanc M, Livingston RB. Survival determinants in extensive-stage non-small cell lung cancer: the Southwest Oncology Group experience. J Clin Oncol 1991; 9: 1618-26. Paesmans M, Sculier JP, Libert P, et al. Prognostic factors for survival in advanced non-small cell lung cancer: univariate and multivariate analyses including recursive partitioning and amalgamation algorithms in 1052 patients. J Clin Oncol 1995; 13: 1221-30.
Shepherd FA. cancer.
2
3
Anthony G Freeman Meadow Rise, 3 Lakeside, Swindon, Wiltshire SN3
1QE, UK
Can antiviral agents decrease the of Kaposi’s sarcoma?
*Considered as time-dependent covariates. Table: Relative hazards (95% CI) and p value for KS in Cox proportional hazards model of time to KS
occurrence
SiR-Following the publication by Chang et all of the identification of herpesvirus-like DNA sequences in AIDSassociated Kaposi’s sarcoma (KS), several teams have reported similar results in The Lancet (March 18, p 722-23; March 25, p 759-61 and p 761-62; April 22, p 1043 and p 1043-44; May 6, p 1180) and elsewhere.2 We studied whether antiviral agents active against herpesvirus could reduce the occurrence of KS, as postulated by Jones et aP and in line with the report by Morfeldt and Torsander.’ The French clinical epidemiology database on HIVseropositive subjects followed in hospitals was started in 1989 in more than 40 hospitals in France. The standardised data collection form includes characterisation of exposure of usual biological markers, clinical group, values manifestations of HIV infection (physician’s diagnosis), nature of treatments delivered, identification of clinical trials in which the patient was included, and death and cause of death as reported in the medical records. A follow-up form is used at each visit or hospital admission at which a clinical manifestation of HIV infection is reported or a new treatment is prescribed, or at least every 6 months. Entry in this study was defined as the time of the first recorded CD4 count. Only patients with a follow-up equal to or longer than 6 months were included. Exclusion criteria were KS at entry or previous reported history of KS and participation in a trial evaluating antiretroviral treatment and/or acyclovir, foscarnet, or ganciclovir. Among the 20 963 subjects included in the database, 16 229 fulfilled the study criteria, 76% were men, and 41 were men who have sex with men. The median follow-up time, from entry to KS or last visit, 20 months (interquartile range 13-33). KS was was diagnosed in 1106 patients at some time during follow-up (7%), and for 48% KS was the first AIDS-defining illness. More than two-thirds of patients (69%) were prescribed antiretroviral treatment before the end of the study period; (16%) were prescribed acyclovir, 782 (5%) ganciclovir, (0-8%) foscarnet. Data were analysed with Cox proportional hazards regression model with time-dependent 2626
and 135
covariates. Results are presented in the table. The relative hazard for KS was slightly increased with acyclovir (1-3 in our study vs 1-4 in Jones et aP); no difference was found with ganciclovir (relative hazard was 1-0 in both studies) or foscarnet (relative hazards 1-4 in our study compared with 0-3 in Jones et al). As in the report by Jones and colleagues, results were similar when the analysis was restricted to men, or to men who have sex with men. Therefore, the results of both studies, which had similar design and analysis, were similar, although the follow-up was longer in our study and fewer medications were prescribed, with the exception of foscarnet for which the two studies are discordant. It should be kept in mind that both studies were observational and therefore potentially subject to 578
comparison bias. In our view, one such bias could explain slight increase in the risk of developing KS after acyclovir. It has been reported that patients who take both zidovudine and acyclovir survive longer. They are therefore more likely to develop disease as their time at risk increases. Could another bias acting in an opposite direction in each study explain the difference? Other evaluations of the relation between Kaposi’s sarcoma and antiviral medications active against herpesviruses are needed to clarify the issue.
the
*D Costagliola, M Mary-Krause, for Clinical Epidemiology Group from Centres d’Information et de Soins de l’Immunodéficience Humaine B3E, INSERM SC4, Institut Fédératif Saint-Antoine de Recherches en Santé, Faculté de Médecine Saint-Antoine, Université Pierre et Marie Curie, 75571 Paris, France
1
2
3 4
5
Chang Y, Cesarman E, Pessin MS, et al. Identification of new human herpesvirus-like DNA sequences in AIDS-associated Kaposi’s sarcoma. Science 1994; 266: 1865-69. Ambroziak JA, Blackbourn DJ, Herndier BG, et al. Herpes-like sequences in HIV-infected and uninfected Kaposi’s sarcoma patients. Science 1995; 268: 582-83. Jones JL, Hanson DL, Chu SY, Ward JW, Jaffe HW. AIDS-associated Kaposi’s sarcoma. Science 1995; 267: 1078-79. Morfeldt L, Torsander J. Long-term remission of Kaposi’s sarcoma following foscarnet treatment in HIV-infected patients. Scand J Infect Dis 1994; 26: 749-52. Cooper DA, Pehrson PO, Pedersen C, et al. The efficacy and safety of zidovudine alone or as cotherapy with acyclovir for the treatment of patients with AIDS and AIDS-related complex: a double-blind randomized trial. AIDS 1993; 7: 197-208.
HIV, occupational exposure, and medical responsibilities management of clinically significant HIV infection requires knowledge of the exposure maximum period between exposure and the development of detectable antibodies in serum. Meyohas and colleagues (June 24, p 1634) suggest that this time could be greater than 6 months, and that current practice should be changed. Before such changes are introduced, the criteria used to define the duration of the seroconversion interval should first be established. These criteria, at the least, should include documentation of (a) a contemporaneously recorded significant exposure to (b) an identifiable source, with the source having (c) laboratory-confirmed evidence of HIV, and where the recipient is (d) shown to be uninfected at the time of the exposure, and (e) shown, through interval testing instigated in response to the reported exposure, to have become infected. The test used to establish the time of the last negative HIV test should be equivalent to that used to determine the time of the first positive test. Reasonable effort should be made to exclude a second, later exposure, including, where relevant, documentation of the HIV status
SiR-Appropriate to
Figure: Periods in which HIV seroconversion took place after occupational exposure 37 percutaneous and four mucocutaneous exposures. Previous
ot sexuai contacts. wnere
cases:
12 in USA, 29 elsewhere.
possiole, Miv 1rom tne putative and the infected recipient should be compared by molecular techniques. These criteria are most likely to be fulfilled after occupational exposure, and review of the 73 cases of definite occupationally acquired HIV infection reported worldwide’I is instructive. In 41 patients, the time of the first positive result of a test for HIV antibody was reported (figure); only in three was anti-HIV first demonstrated at more than 6 months after exposure. One was detected at blood donation 9 months after mucocutaneous exposure, having tested negative at 8 weeks post exposure; a second exposure, to a phlebotomy needle used on an injecting drug user of unknown HIV status, happened 2 months after the first.2 The second case was detected when contacted for testing with a new anti-HIV test in 1986, 11 months after percutaneous exposure to blood from an injecting drug user who was anti-HIV negative but had detectable p24 antigen.3 In the third, seroconversion took place between day 90 and day 288.4 None of the three recalled a seroconversion illness. One had tested anti-HIV negative at 3 months post exposure, but none had been tested at 6 months. 29 (63%) of 46 cases described in detail had symptoms compatible with acute HIV infection. Most symptoms (in 26, 90%) had arisen between the 2nd and 6th weeks after exposure. One illness occurred more than 3 months later: after exposure in November, 1985, the worker tested anti-HIV negative for 3 months, developed symptoms in April, 1986, and was proven infected in May, 1986. A second exposure was not excluded.5 Meyohas and colleagues do not report the infection source in their patient, and the combination of late onset of possible symptoms and late seroconversion is unusual (figure). Further details about the exposure, the source, and the process by which a later exposure was excluded would help in assessment of the public health importance of this event. The transmission rate after a single percutaneous exposure to a known HIV infected source has been estimated as being around 1 in 300: most of those so exposed will not become infected. Achieving compliance with anti-HIV testing even at baseline, 6 weeks, 3 months, and 6 months post exposure is difficult. We believe that there is insufficient evidence to justify change in what has inevitably become, despite source
guidance recommending testing exposure, common practice. *Julia
at
12
months
after
Heptonstall, O Noel Gill
PHLS Communicable Disease Surveillance Centre, London NW9 5EQ, UK
1
Heptonstall J, Porter KP, Gill ON. Occupational transmission of HIV: summary of published reports to July 1995. London: Public Health 1995 (internal report). Centers for Disease Control. Update: human immunodeficiency virus infections in health care workers exposed to blood of infected patients. MMWR 1987; 36: 285-89. Ippolito G, Puro V, de Carli G, and the Italian Study Group on Occupational Risk of HIV Infection. The risk of occupational human immunodeficiency virus infection in health care workers. Arch Int Med 1993; 153: 1451-58. Ramsey KM, Smith EN, Reinarz JA. Prospective evaluation of 44 health care workers exposed to human immunodeficiency virus-1, with one seroconversion. Clin Res 1988; 36: 22A. Wallace MR, Harrison WO. HIV seroconversion with progressive disease in health care worker after needlestick injury. Lancet 1988; i: 1454.
Laboratory Service, 2
3
4
5
and colleagues’ report of HIV seroconversion after needlestick injury in a woman deserves comment. Because medical information on HIV and AIDS is largely disseminated by the media, we should be prudent before publishing such clinical cases. No rigorous scientific conclusion can be drawn from Meyohas’ report because the presumed absence of risk factors for HIV, other than needlestick injury, between May and November, 1993, is based solely on the reliability of questioning. That the needle was used in an HIV-infected patient is not certain, and thus no information is available on the patient’s clinical and biological status. Comparison of nucleotide sequence between HIV isolates from the patient and the woman would have provided true evidence for a connection between the needlestick injury and the syndrome of primary HIV infection, as it has been in other circumstances.’ Without these data it is very hazardous to drawn public health conclusions from this observation, as Meyohas and colleagues have done. In France, this case report was publicised 2 days before we received the issue of The Lancet concerned. This coverage caused much anxiety in nurses who had been accidentally exposed to HIV in past months. Finally, we know that the time between submission of and
SIR-Meyohas
579
letter is short, so we do not understand why this case was not reported at the end of 1993, especially since it is of such public health importance.
publication of a
*Alain Lafeuillade, Catherine Thiebaut, Robert Quilichini, Nérina Profizi *Department of Infectious Diseases and Laboratory of Virology General Hospital, 83056 Toulon, France
1
Jaffe HW, McCurdy JM, Kalish ML,
et al. Lack of HIV transmission in the practice of a dentist with AIDS. Ann Intern Med 1994; 121: 855-59.
authors’
reply
SiR-In
Heptonstall’s report, a small number of patients negative for anti-HIV antibody beyond the third month after exposure to HIV. As in our findings, the existence of a delayed preseroconversion period is difficult to establish formally. However, it cannot be excluded in rare cases. Our only aim, when publishing our findings in The Lancet, was to ask of the scientific community whether other similar cases had been observed worldwide. We reported this observation only for the sake of vigilance in public health. Lafeuillade raises the issue of worrying the general population with such publication. The cases of HIV-0 infection, reported in The Lancet several months ago, could also worry the general population in so far as they might be an underestimate of the diagnosis of infection. However, the publication of our findings was useful. Our observation appeared sufficiently troubling to report to the scientific community. We chose to wait until 1995 before publishing our findings, to ensure through long follow-up and continued interviews that the patient did not have other risk factors of HIV infection during the critical period.
were
M C
Meyohas,
L
Morand-Joubert, *J J Lefrère
INTS Hôpital St-Antoine, 75012 Paris, France
Table: Multivariate
Chemotherapy for elderly patients with lung cancer
SiR-More than 50% of patients with inoperable non-small cell lung cancer (NSCLC) are over the age of 65, yet palliative chemotherapy, if given at all, is usually reserved for younger age groups. Are there any sound data for denying these elderly patients treatment? The upper age limit in only one of eight authoritative randomised trials of chemotherapy versus supportive care for patients with advanced NSCLC extended to 75 years, and the average of the median ages in this group of trials was about 60 years (reviewed in ref 1). There was no evidence from these trials that the elderly fared less well with chemotherapy. The impact of age as a prognostic factor for survival and response in advanced NSCLC has been analysed in reports totalling 4516 patients2.3 (and reviewed in ref 3). In one of these, the SouthWest Oncology Group analysed a database of 2531 patients with stage IV NSCLC2 and found by multivariate analysis that age greater than 70 years was in fact a significant independent positive predictor for survival (p=002). In the remainder of these studies age had no influence by multivariate analysis. Since 1990 we have prospectively generated a database of patients with a proven diagnosis of inoperable NSCLC, referred to the lung unit at the Royal Marsden Hospital. We have now analysed 39 pretreatment variables for their influence on objective response, symptomatic response, and survival to chemotherapy (usually cisplatin-based) in 290 patients treated in a series of phase II trials. The table lists 580
factors identified
analysis for influential prognostic factors
by multivariate analysis
with the stepwise method. Consistent with other studies/,3
logistic regression performance status (Eastern Cooperative Oncology Group),
disease extent, and certain biochemical features were influential variables for survival and objective response, and we also show that this holds for symptom relief-a crucial outcome for treatment given with palliative intent. Additionally, older age was positively associated with objective response. Age had no influence, either positive or negative, on symptom relief or survival. These data indicate that older patients fare just as well as younger ones with palliative chemotherapy, and possibly better. Given the age distribution of NSCLC and the growing elderly population, palliative treatment of the elderly with this disease will increasingly be an issue. We consider that elderly patients with advanced NSCLC should no longer be excluded from palliative chemotherapy trials. T F Hickish, *I E Smith, S Ashley, G Middleton Lung Unit, Royal Marsden NHS Trust, Sutton, Surrey SM2 5PT, UK
1
Future directions in the treatment of non-small cell lung Semin Oncol 1994; 21: 48-62. Albain KS, Crowley JJ, LeBlanc M, Livingston RB. Survival determinants in extensive-stage non-small cell lung cancer: the Southwest Oncology Group experience. J Clin Oncol 1991; 9: 1618-26. Paesmans M, Sculier JP, Libert P, et al. Prognostic factors for survival in advanced non-small cell lung cancer: univariate and multivariate analyses including recursive partitioning and amalgamation algorithms in 1052 patients. J Clin Oncol 1995; 13: 1221-30.
Shepherd FA. cancer.
2
3