- Email: [email protected]
Fertility drugs and breast and ovarian cancer
upregulated in psoriasis. Atopic eczema was differentiated by a combination of FceRI and FcyRII analysis (figure). Langerhans-cell phenotyping may identify subtle differences between inflammatory skin lesions of different origins; a strategy that represents a helpful tool in the differential diagnosis of inflammatory skin lesions. *Andreas
Wollenberg, SongPing Wen,
Thomas Bieber
Department of Dermatology, Ludwig-Maximilians-University, D-80337 Munich, Germany
Hanifin J, Rajka G. Diagnostic features of atopic dermatitis. Acta Derm Venerol 1980; 92: 44-47. 2 Bos J, Kapsenberg ML, Smitt JH. Pathogenesis of atopic eczema. Lancet 1994; 343: 1338-41. 3 Wollenberg A, Kraft S, Hanau D, Bieber T. Immunomorphological and ultrastructural characterization of Langerhans cells and a novel, inflammatory dendritic epidermal cell (IDEC) population in lesional skin of atopic eczema. J Invest Dermatol (in press). 1
Treatment of
inoperable ovarian interleukin-2 and carboplatinum
cancer
with
discovery of interleukin-2 (11-2), a cytokine produced by T lymphocytes with in-vivo cytotoxic activity against cancer cells has focused attention on antineoplastic immunotherapy.’ 11-2 is used in the treatment of renal cancer and melanoma2 and has induced complete, longterm remission, not achieved with traditional therapy. A 67-year-old woman came to us for treatment four months ago. She had striking ascites and a hard swelling in the left pelvis. Her performance status was 60 on the Kamofsky scale. Ultrasonography and computed tomography confirmed a round tumour with irregular margins 2-2 cm in diameter in the region of the left ovary, compatible with cystoadenocarcinoma. After paracentesis and intraperitoneal administration of 18 million U of 11-2, her condition improved and she was then given carboplatinum (300 mg/month, dose calculated on the basis of creatinine clearance) for 4 months, with 11-2 (4-5 million U/day, subcutaneously for 4 days). Computed tomography after the fourth chemotherapy cycle showed striking reduction in size of the cancer. She had a total hysterectomy with pelvic and lumboaortic lymphoadenectomy. Histological examination revealed bilateral serous pT3a NO MO ovarian papillary cystoadenocarcinoma in the form of microscopic residues in the ovarian cortex. She remains well with a Kamofsky score of 90. In our case, 11-2 did not cause side-effects3 nor provoke peritoneal damage. Its cytotoxic effect may have been due to uniform distribution throughout the abdominal cavity because of the ascites. Although 11-2 needs to be used with SIR-The
prudence, this
case
confirms the usefulness of combined
chemotherapy in solid tumours where other chemotherapeutic drugs would not have been successful.
treatment with
*Mariano Malaguarnera, Paolo Scollo, Roberto Ruggeri, Sebastiano Caudullo, Giovanni Pistone *Institutes of Internal Medicine and Geriatrics, University of Catania, 95126 Catonia, Italy; Institute of Gynecology and Obstetrics, Ospedale Cannizzaro, Catania
1
2
3
Lotze MT, Muul LM. Observation of the systemic administration of autologus lymphokine activated killer cells and recombinant interleukin-2 to patients with cancer. N Engl J Med 1985; 313: 1485-92. Rosenberg SA. Immunotherapy of patient with advanced cancer, using interleukin-2 alone or in combination with lymphokine-activated killer cells. In: DeVita VT, Hellman S, Rosenberg SA, eds. Recent advances in oncology. Philadelphia: JB Lippincott Co, 1988: 217-57. Lee RE, Lotze MT, Skibber JM. Cardiorespiratory effects of immunotherapy with interleukin 2. J Clin Oncol 1989; 7: 7-20.
Rosenberg SA,
Fertility drugs
and breast and ovarian
cancer
SIR-Venn and colleagues (Oct 14, p 995) examine the risk of invasive ovarian and breast cancer in women referred to a specialty clinic for in-vitro fertilisation (IVF). Women who registered for IVF but did not go on to receive this treatment were followed for cancer incidence from the date of registration, whereas women who underwent one or more stimulated IVF procedures were followed from the date of their first treatment cycle. Omitted from the analysis were the months-to-years that accrued between the time of registration and the time of first IVF stimulation in women who received IVF. If a cancer was diagnosed during this interval, it was tabulated as having occurred in a woman who had not undergone IVF. By failing to include the person-time before the start of IVF in the calculation of cancer rates in so-called unexposed women, Venn and colleagues have overestimated these rates, and thus underestimated the relative rates in women who did receive IVF. Whether Venn’s results would be altered to an important extent if they were to calculate the person-time experience of the study population, as we suggest, is not clear. However, the diagnosis of ovarian cancer in one untreated woman 1-6 years after registration, in combination with these workers’ statement that the waiting time for IVF treatment was as long as 3 years for some women, suggests that such an analysis should be undertaken.
*Mary Anne Rossing,
Noel S Weiss
Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA 98104, USA
Authors’
reply
SIR-Rossing and Weiss allude to an inconsistency in our analysis of women exposed and unexposed to IVF treatment and suggest how this may have contributed to an overestimation of the rate of cancer in the unexposed group. We acknowledge that it would have been preferable to use time of registration for all women and to have recorded the time of first IVF treatment as an additional data point. In the short-term it is not feasible to return to the archived files to incorporate the date of registration for women in the exposed group, which was not always recorded or retained once treatment had started. Contrary to the assumption made by Rossing and Weiss, however, we had not included in our standardised incidence ratio (SIR) and relative risk (RR) estimates any cases of cancer that had occurred in the exposed group before their first stimulated cycle. To attempt to answer the query about the effect of including person-time between registration and first stimulated cycle we include here a sensitivity analysis whereby we incorporate into the dataset a best estimate of time of registration for the exposed group. This analysis is based on the understanding of waiting times by one of us (C K) involved with the cohort throughout its existence. Hence, included was 3 years of waiting time for those having treatment before 1983, 1 year between 1983 and 1985, 9 months between 1985 and 1988, and 3 months thereafter. This inclusion led to an additional 4870 person-years being incorporated into the unexposed total and 1-36, 0-17, and 0-06 being added to the expected numbers for breast, ovarian, and body of uterus (uterine) cancers, respectively. Since there were no invasive cancers in the exposed group occurring before IVF treatment in this estimated waiting period, the observed numbers of cancers did not change. The table shows the new SIRs for the unexposed group for breast, ovarian, and uterine cancers. These new estimates do not alter the conclusions drawn earlier. 1627
Obs=observed,Exp=expected. Table: SIRs for breast, ovarian, and uterine
Use of
cancer
according to fertility treatment
hazards model and a timedependent covariate for exposure provided an RR estimate for breast cancer of 1-09 (95% CI 0-55-2-16). However, with such small numbers of ovarian and uterine cancer cases, this model proved troublesome, giving unstable RR estimates for these cancers with the additional person-years data in the model. Longer-term follow-up of a larger cohort is needed to obtain more reliable RR estimates for these rarer gynaecological cancers. We shall certainly include registration date in our further multicentre study which will start in 1996. a
proportional
*Alison Venn, Lyndsey Watson, Judith Catriona King, David Healy
Lumley,
Graham
Centre for the Study of Mothers’ and Children’s Health, La Trobe Victoria 3053, Australia
,
Giles,
University, Carlton,
SiR-Venn and colleagues (Oct 14, p 995), on the basis of a cohort of women referred for in-vitro fertilisation, provide reassuring data on the possible relation between fertility drugs and ovarian cancer. The issue has attracted lively interest’ following an overview of American case-control studies,2 which showed an increased ovarian cancer risk among women who had used fertility drugs, particularly for tumours of low malignant potential,3 and among nulliparae. An increased risk of borderline neoplasms was also reported in a cohort of infertile women from Washington stated Our group has considered data from a case-control study conducted in Milan, part of these data being included in a cooperative Italian study.’ A total of 208 patients (aged 18-75, median 54) with histologically confirmed epithelial ovarian cancer, and 873 non-gynaecological, non-malignant controls (aged 19-75, median 56) were interviewed in Milan between 1992 and June, 1995. Only 4 (1-9%) of the 208 women with ovarian cancer, and 13 (1-5%) control group patients reported ever using fertility drugs; this corresponds to a multivariate odds ratio of 1-1 (95% CI 0-4-3-6). None of 34 nulliparous ovarian cancer patients versus two of 135 nulliparous controls had used fertility drugs. Our data provide further reassurance about the absence of a strong and consistent association between exposure to fertility drugs and subsequent risk of developing epithelial ovarian cancer. Conducted within the framework of the Italian National Research Council FATMA and ACRO Projects (Contracts No.95.00952.PF41 and
94.01321.PF29). *Carlo La Vecchia, Eva
Negri,
Fabio Parazzini, Silvia Franceschi
*Istituto di Ricerche Farmacologiche "Mario Negri", 20157 Milan, Italy; Istituto di Statistica Medica e Biometria, Università degli Studi di Milano, Milan, I Clinica Ostetrico-Ginecologica, Università degli Studi di Milano, Milan; and Centro di Riferimento Oncologico, Aviano
1 2
3
Balasch J, Barri PN. Follicular stimulation and ovarian cancer? Human Reprod 1993; 8: 990-96. Whittemore AS, Harris R, Itnyre J, and the Collaborative Ovarian Cancer Group. Characteristics relating to ovarian cancer risk: collaborative analysis of 12 case-control studies II, invasive epithelial ovarian cancers in white women. Am J Epidemiol 1992; 136: 1184-203. Harris R, Whittemore AS, Itnyre J, and the Collaborative Ovarian Cancer Group. Characteristics relating to ovarian cancer risk: collaborative analysis of 12 US case-control studies III, epithelial tumors of low malignant potential in white women. Am J Epidemiol
1992; 136: 1204-11.
1628
4
Moore DE, Self SG. Ovarian in a cohort of infertile women. N Engl J Med 1994; 331: 771-76. Franceschi F, La Vecchia C, Negri E, et al. Fertility drugs and risk of epithelial ovarian cancer in Italy. Hum Reprod 1994; 9: 1673-75.
Rossing MA, Daling JR, Weiss NS, tumors
5
Identifying areas
at
high-risk for neonatal
tetanus SIR-Immunisation of pregnant women and women of childbearing age with tetanus toxoid (TT) is the main control measure for neonatal tetanus (NNT), and the global plan for elimination of NNT emphasises the need to focus such immunisation activities on high risk areas.’ Household-based surveys of maternal and child health, and NNT, were undertaken in 1993 and 1994 in 282 counties in eight Provinces of the People’s Republic of China. Surveys of 31 counties in Gansu Province along with reported NNT deaths from each county for 1991, 1992, and 1993 were made available by Gansu Province Health authorities. TT coverage in women of childbearing age was below 4% in 28 of the counties, and about 24% in three counties that had undertaken immunisation campaigns. Given high case-fatality ratios, NNT cases and deaths should be similar. Reports were available in all 3 years for 22 counties, but were missing in 1 of the 3 reporting years for eight counties, and one county had reported only once. 13 counties in Gansu Province had NNT rates of 4 or more/1000 live births but only three had reported NNT deaths and live births in 1993 consistent with such NNT rates, assuming case-fatality ratios of 50% or more. Underreporting of NNT cases and deaths from routine reporting systems is widely recognised. Two operational criteria to identify high-risk areas using reported cases have been proposed by the Pan-American Health Organisation: (1) reports of one or more cases of NNT in each of the past 3 years; and (2) reports of one or more cases from an area not previously identified as high risk. The data from Gansu Province provided an opportunity to test such an assessment, with NNT defined by deaths rather by cases. (1) (above) had a low sensitivity, 54% (7/13), but a high specificity, 83% (15/18), in identifying first priority counties within Gansu Province. (2) (above) is equivalent to reports of any cases at any time during the 3-year period since all counties were deemed at unknown risk before 1991. 12 truly highest risk counties were positive by this criterion, giving a high sensitivity, 92% (12/13), but lower specificity, 28% (5/18). With adequate resources, such outcomes should be acceptable in selecting counties that will be given special priority for campaigns with TT. However, the efficiency of selection is low, since most (13 of the 25) counties selected for campaigns are not first priority. A report of no cases in any of the 3 years proved meaningful, since it correctly identified 14 of the 18 lower risk counties, while incorrectly identifying four of 13 first priority counties. A combination of reports of any cases at anytime and the absence of no reports during the 3 reporting years can be used to select high-priority counties; this criterion has a sensitivity of 69% (9/13), specificity 78% (14/18), and improves the efficiency of selection with nine of 13 (69%) selected counties being
Wollenberg, SongPing Wen,
Thomas Bieber
Department of Dermatology, Ludwig-Maximilians-University, D-80337 Munich, Germany
Hanifin J, Rajka G. Diagnostic features of atopic dermatitis. Acta Derm Venerol 1980; 92: 44-47. 2 Bos J, Kapsenberg ML, Smitt JH. Pathogenesis of atopic eczema. Lancet 1994; 343: 1338-41. 3 Wollenberg A, Kraft S, Hanau D, Bieber T. Immunomorphological and ultrastructural characterization of Langerhans cells and a novel, inflammatory dendritic epidermal cell (IDEC) population in lesional skin of atopic eczema. J Invest Dermatol (in press). 1
Treatment of
inoperable ovarian interleukin-2 and carboplatinum
cancer
with
discovery of interleukin-2 (11-2), a cytokine produced by T lymphocytes with in-vivo cytotoxic activity against cancer cells has focused attention on antineoplastic immunotherapy.’ 11-2 is used in the treatment of renal cancer and melanoma2 and has induced complete, longterm remission, not achieved with traditional therapy. A 67-year-old woman came to us for treatment four months ago. She had striking ascites and a hard swelling in the left pelvis. Her performance status was 60 on the Kamofsky scale. Ultrasonography and computed tomography confirmed a round tumour with irregular margins 2-2 cm in diameter in the region of the left ovary, compatible with cystoadenocarcinoma. After paracentesis and intraperitoneal administration of 18 million U of 11-2, her condition improved and she was then given carboplatinum (300 mg/month, dose calculated on the basis of creatinine clearance) for 4 months, with 11-2 (4-5 million U/day, subcutaneously for 4 days). Computed tomography after the fourth chemotherapy cycle showed striking reduction in size of the cancer. She had a total hysterectomy with pelvic and lumboaortic lymphoadenectomy. Histological examination revealed bilateral serous pT3a NO MO ovarian papillary cystoadenocarcinoma in the form of microscopic residues in the ovarian cortex. She remains well with a Kamofsky score of 90. In our case, 11-2 did not cause side-effects3 nor provoke peritoneal damage. Its cytotoxic effect may have been due to uniform distribution throughout the abdominal cavity because of the ascites. Although 11-2 needs to be used with SIR-The
prudence, this
case
confirms the usefulness of combined
chemotherapy in solid tumours where other chemotherapeutic drugs would not have been successful.
treatment with
*Mariano Malaguarnera, Paolo Scollo, Roberto Ruggeri, Sebastiano Caudullo, Giovanni Pistone *Institutes of Internal Medicine and Geriatrics, University of Catania, 95126 Catonia, Italy; Institute of Gynecology and Obstetrics, Ospedale Cannizzaro, Catania
1
2
3
Lotze MT, Muul LM. Observation of the systemic administration of autologus lymphokine activated killer cells and recombinant interleukin-2 to patients with cancer. N Engl J Med 1985; 313: 1485-92. Rosenberg SA. Immunotherapy of patient with advanced cancer, using interleukin-2 alone or in combination with lymphokine-activated killer cells. In: DeVita VT, Hellman S, Rosenberg SA, eds. Recent advances in oncology. Philadelphia: JB Lippincott Co, 1988: 217-57. Lee RE, Lotze MT, Skibber JM. Cardiorespiratory effects of immunotherapy with interleukin 2. J Clin Oncol 1989; 7: 7-20.
Rosenberg SA,
Fertility drugs
and breast and ovarian
cancer
SIR-Venn and colleagues (Oct 14, p 995) examine the risk of invasive ovarian and breast cancer in women referred to a specialty clinic for in-vitro fertilisation (IVF). Women who registered for IVF but did not go on to receive this treatment were followed for cancer incidence from the date of registration, whereas women who underwent one or more stimulated IVF procedures were followed from the date of their first treatment cycle. Omitted from the analysis were the months-to-years that accrued between the time of registration and the time of first IVF stimulation in women who received IVF. If a cancer was diagnosed during this interval, it was tabulated as having occurred in a woman who had not undergone IVF. By failing to include the person-time before the start of IVF in the calculation of cancer rates in so-called unexposed women, Venn and colleagues have overestimated these rates, and thus underestimated the relative rates in women who did receive IVF. Whether Venn’s results would be altered to an important extent if they were to calculate the person-time experience of the study population, as we suggest, is not clear. However, the diagnosis of ovarian cancer in one untreated woman 1-6 years after registration, in combination with these workers’ statement that the waiting time for IVF treatment was as long as 3 years for some women, suggests that such an analysis should be undertaken.
*Mary Anne Rossing,
Noel S Weiss
Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA 98104, USA
Authors’
reply
SIR-Rossing and Weiss allude to an inconsistency in our analysis of women exposed and unexposed to IVF treatment and suggest how this may have contributed to an overestimation of the rate of cancer in the unexposed group. We acknowledge that it would have been preferable to use time of registration for all women and to have recorded the time of first IVF treatment as an additional data point. In the short-term it is not feasible to return to the archived files to incorporate the date of registration for women in the exposed group, which was not always recorded or retained once treatment had started. Contrary to the assumption made by Rossing and Weiss, however, we had not included in our standardised incidence ratio (SIR) and relative risk (RR) estimates any cases of cancer that had occurred in the exposed group before their first stimulated cycle. To attempt to answer the query about the effect of including person-time between registration and first stimulated cycle we include here a sensitivity analysis whereby we incorporate into the dataset a best estimate of time of registration for the exposed group. This analysis is based on the understanding of waiting times by one of us (C K) involved with the cohort throughout its existence. Hence, included was 3 years of waiting time for those having treatment before 1983, 1 year between 1983 and 1985, 9 months between 1985 and 1988, and 3 months thereafter. This inclusion led to an additional 4870 person-years being incorporated into the unexposed total and 1-36, 0-17, and 0-06 being added to the expected numbers for breast, ovarian, and body of uterus (uterine) cancers, respectively. Since there were no invasive cancers in the exposed group occurring before IVF treatment in this estimated waiting period, the observed numbers of cancers did not change. The table shows the new SIRs for the unexposed group for breast, ovarian, and uterine cancers. These new estimates do not alter the conclusions drawn earlier. 1627
Obs=observed,Exp=expected. Table: SIRs for breast, ovarian, and uterine
Use of
cancer
according to fertility treatment
hazards model and a timedependent covariate for exposure provided an RR estimate for breast cancer of 1-09 (95% CI 0-55-2-16). However, with such small numbers of ovarian and uterine cancer cases, this model proved troublesome, giving unstable RR estimates for these cancers with the additional person-years data in the model. Longer-term follow-up of a larger cohort is needed to obtain more reliable RR estimates for these rarer gynaecological cancers. We shall certainly include registration date in our further multicentre study which will start in 1996. a
proportional
*Alison Venn, Lyndsey Watson, Judith Catriona King, David Healy
Lumley,
Graham
Centre for the Study of Mothers’ and Children’s Health, La Trobe Victoria 3053, Australia
,
Giles,
University, Carlton,
SiR-Venn and colleagues (Oct 14, p 995), on the basis of a cohort of women referred for in-vitro fertilisation, provide reassuring data on the possible relation between fertility drugs and ovarian cancer. The issue has attracted lively interest’ following an overview of American case-control studies,2 which showed an increased ovarian cancer risk among women who had used fertility drugs, particularly for tumours of low malignant potential,3 and among nulliparae. An increased risk of borderline neoplasms was also reported in a cohort of infertile women from Washington stated Our group has considered data from a case-control study conducted in Milan, part of these data being included in a cooperative Italian study.’ A total of 208 patients (aged 18-75, median 54) with histologically confirmed epithelial ovarian cancer, and 873 non-gynaecological, non-malignant controls (aged 19-75, median 56) were interviewed in Milan between 1992 and June, 1995. Only 4 (1-9%) of the 208 women with ovarian cancer, and 13 (1-5%) control group patients reported ever using fertility drugs; this corresponds to a multivariate odds ratio of 1-1 (95% CI 0-4-3-6). None of 34 nulliparous ovarian cancer patients versus two of 135 nulliparous controls had used fertility drugs. Our data provide further reassurance about the absence of a strong and consistent association between exposure to fertility drugs and subsequent risk of developing epithelial ovarian cancer. Conducted within the framework of the Italian National Research Council FATMA and ACRO Projects (Contracts No.95.00952.PF41 and
94.01321.PF29). *Carlo La Vecchia, Eva
Negri,
Fabio Parazzini, Silvia Franceschi
*Istituto di Ricerche Farmacologiche "Mario Negri", 20157 Milan, Italy; Istituto di Statistica Medica e Biometria, Università degli Studi di Milano, Milan, I Clinica Ostetrico-Ginecologica, Università degli Studi di Milano, Milan; and Centro di Riferimento Oncologico, Aviano
1 2
3
Balasch J, Barri PN. Follicular stimulation and ovarian cancer? Human Reprod 1993; 8: 990-96. Whittemore AS, Harris R, Itnyre J, and the Collaborative Ovarian Cancer Group. Characteristics relating to ovarian cancer risk: collaborative analysis of 12 case-control studies II, invasive epithelial ovarian cancers in white women. Am J Epidemiol 1992; 136: 1184-203. Harris R, Whittemore AS, Itnyre J, and the Collaborative Ovarian Cancer Group. Characteristics relating to ovarian cancer risk: collaborative analysis of 12 US case-control studies III, epithelial tumors of low malignant potential in white women. Am J Epidemiol
1992; 136: 1204-11.
1628
4
Moore DE, Self SG. Ovarian in a cohort of infertile women. N Engl J Med 1994; 331: 771-76. Franceschi F, La Vecchia C, Negri E, et al. Fertility drugs and risk of epithelial ovarian cancer in Italy. Hum Reprod 1994; 9: 1673-75.
Rossing MA, Daling JR, Weiss NS, tumors
5
Identifying areas
at
high-risk for neonatal
tetanus SIR-Immunisation of pregnant women and women of childbearing age with tetanus toxoid (TT) is the main control measure for neonatal tetanus (NNT), and the global plan for elimination of NNT emphasises the need to focus such immunisation activities on high risk areas.’ Household-based surveys of maternal and child health, and NNT, were undertaken in 1993 and 1994 in 282 counties in eight Provinces of the People’s Republic of China. Surveys of 31 counties in Gansu Province along with reported NNT deaths from each county for 1991, 1992, and 1993 were made available by Gansu Province Health authorities. TT coverage in women of childbearing age was below 4% in 28 of the counties, and about 24% in three counties that had undertaken immunisation campaigns. Given high case-fatality ratios, NNT cases and deaths should be similar. Reports were available in all 3 years for 22 counties, but were missing in 1 of the 3 reporting years for eight counties, and one county had reported only once. 13 counties in Gansu Province had NNT rates of 4 or more/1000 live births but only three had reported NNT deaths and live births in 1993 consistent with such NNT rates, assuming case-fatality ratios of 50% or more. Underreporting of NNT cases and deaths from routine reporting systems is widely recognised. Two operational criteria to identify high-risk areas using reported cases have been proposed by the Pan-American Health Organisation: (1) reports of one or more cases of NNT in each of the past 3 years; and (2) reports of one or more cases from an area not previously identified as high risk. The data from Gansu Province provided an opportunity to test such an assessment, with NNT defined by deaths rather by cases. (1) (above) had a low sensitivity, 54% (7/13), but a high specificity, 83% (15/18), in identifying first priority counties within Gansu Province. (2) (above) is equivalent to reports of any cases at any time during the 3-year period since all counties were deemed at unknown risk before 1991. 12 truly highest risk counties were positive by this criterion, giving a high sensitivity, 92% (12/13), but lower specificity, 28% (5/18). With adequate resources, such outcomes should be acceptable in selecting counties that will be given special priority for campaigns with TT. However, the efficiency of selection is low, since most (13 of the 25) counties selected for campaigns are not first priority. A report of no cases in any of the 3 years proved meaningful, since it correctly identified 14 of the 18 lower risk counties, while incorrectly identifying four of 13 first priority counties. A combination of reports of any cases at anytime and the absence of no reports during the 3 reporting years can be used to select high-priority counties; this criterion has a sensitivity of 69% (9/13), specificity 78% (14/18), and improves the efficiency of selection with nine of 13 (69%) selected counties being