- Email: [email protected]
CHILDHOOD LEUKAEMIA IN WEST BERKSHIRE
212 CHILDHOOD LEUKAEMIA IN WEST BERKSHIRE
SIR,-The letter by Dr Barton and colleagues (Nov 30, p 1248) is a timely reminder that the distribution of childhood leukaemia (and of many related conditions in adults) is heterogeneous in space. 1984 data for. childhood leukaemia (ICD 204-8), from the Leukaemia Research Fund 1984 Data Collection Survey, show that rates within one year vary up to two-fold between different health regions (table). Although none of the observed/expected differences in the table are significant at the 5% level, the ratio for South West England is similar to that reported from West Berkshire. Other spatial data 1984 CHILDHOOD ACUTE LEUKAEMIA (AGE 0-9) IN SELECTED REGIONS OF 1
BRITAIN-r 1
I
incomplete to a degree ranging from 8% to 32%. The degree of completeness of other regional cancer registries is unknown. To make comparisons between the incidence of childhood leukaemia in a district with special data collection and national figures obtained routinely is likely to be misleading. In five districts in Wessex for which additional data are available supplement cancer registry data the average incidence of acute lymphoblastic leukaemia in children aged 0-9 years for the years 1972-82 ranged between 5-2and 6-4per 100000. The denominator used in the calculations was 1981 census data. A study of the geographical distribution of childhood leukaemia in Dorset shows a gradient of incidence from low rates in urban centres through higher rates in peripheral/commuter communities to higher rates still in rural areas, and a similar gradient from urban centres outwards has been noted elsewhere in Wessex (a detailed report is in preparation). This indicates that even where comprehensive data are available the balance between urban and rural communities must be taken into account when comparisons are being made. to
Central Administrative Offices, Royal Victoria Hospital, Bournemouth BH1 4HX
R. BARCLAY
SPLENECTOMY BEFORE BONE MARROW TRANSPLANTATION IN CHRONIC GRANULOCYTIC LEUKAEMIA
*Including smaller areas which are not detailed in table. tComputed on basis of 1981 census population data.
available in Yorkshire over a longer period of time suggest that variation is commonly seen in most leukaemia/lymphomas. When smaller areas are examined from the 1984 survey significant differences do emerge between expected and observed numbers, often for small numbers of cases in areas with tiny populations. Results highlighting specific areas are of interest but are usually uninterpretable because they are post hoc observations or contain unusual and biologically mixed subgroups of cases and do not account for the generality of cases found by a similar method of case ascertainment. This is one advantage of the 1984 Data Collection Survey which is currently being analysed with a view to understanding the true nature of the leukaemias and related conditions throughout certain parts of the UK. Uniform case ascertainment and analysis have been achieved without collection bias. Even so, cases very close together in space and time have been recorded from several localities in 1984 and 1985. Some clusters, after age and sex standardisation, are very unlikely to be due to chance and these are being fully investigated by the Leukaemia Research Fund cluster investigation group. The scientific community is not ignoring the investigation of such close case occurrences. Clusters need to be investigated responsibly, taking into account all known aetiological factors and not just those that reflect the prejudices of the media. Doctors who think that they have such case occurrence" worth investigating should contact me. I thank the many paediatric oncologists, haematologists, and histopathhave provided data for the Leukaemia Research Fund study.
ologists who
Leukaemia Research Fund 1984/85
Cookridge Hospital, Leeds LS16 6QB
Study, R. A. CARTWRIGHT
SiR,-We read with interest the paper by Dr Gratwohl and colleagues (Dec 7, p 1290) about the effect of splenectomy before bone-marrow transplantation (BMT) on survival in chronic granulocytic leukaemia (CGL). However, some of their conclusions ("debulking in this disease is of no value", "neither splenectomy nor splenic irradiation altered survival") seem premature and are not supported by statistical analysis. There are serious methodological objections to this multi centre non-randomised
study. Gratwohl et al compared a splenectomy group with a no splenectomy group. This comparison is not appropriate because splenectomised patients constitute a heterogeneous group (some with splenic irradiation and some without), and splenic irradiation may be more debulking than splenectomy. To evaluate the effect of splenectomy, it would have been better to have compared the splenectomy group with a no splenectomy/no irradiation group. Other comparisons to estimate the contribution to survival of an "asplenic status" (splenectomy plus splenic irradiation vs no splenectomy and no further therapy group) and of splenic irradiation (irradiation vs no irradiation/no splenectomy) would have been interesting. The second comparison Gratwohl et al did do-a simultaneous comparison of the three groups-does seem to have been done with heterogeneity in mind. If splenic irradiation may be more debulking than splenectomy, a test for trend’ would have been better, especially since the actuarial survival and the shape of the survival curves showed apparently ordered groups of risk (7lTo probability of survival in irradiated patients, 58% in splenectomised patients, and 49% in non-splenectomised, nonirradiated patients). These results must anyway be interpreted with caution because of the small numbers of patients at long-term risk in each category. The impact of splenectomy or splenic irradiation on survival in CGL patients treated by BMT must be investigated in prospective rar.domised trials before routine splenectomy is abandoned. Clinical
Haematology Section,
Haematology/Haemotherapy Service,
SIR,-Dr Barton and her colleagues report an average incidence of leukaemia in children between the ages of 0-9 years for the years 1972-1984 in West Berkshire of 5-88 per 100 000, compared with an average for England and Wales for the years 1972-81 of 4 - 3 per 100 000. In several districts in the Wessex region data about childhood leukaemia have been collected from a variety of sources, including the Wessex Cancer Registry, and paediatric, haematology and community child health departments, and these indicate that for acute lymphoblastic leukaemia the Wessex Cancer Registry is
"La Fe", Valencia, Spain
Hospital 46009
1 Peto R, Pike
requiring
MIGUEL A. SANZ GUILLERMO SANZ IGNACIO LORENZO
MC, Armitage P, et al. Design and analysis of randomized clinical trials prolonged observation of each patient. Br J Cancer 1977, 35: 1-39
SIR,-The report from the European Cooperative Group for Bone Marrow Transplantation provides important conclusions about the role of splenectomy in patients with CGL who are subsequently given a bone marrow transplant (BMT). Splenectomy was of no benefit in terms of survival, prevention of relapse, or incidence of
SIR,-The letter by Dr Barton and colleagues (Nov 30, p 1248) is a timely reminder that the distribution of childhood leukaemia (and of many related conditions in adults) is heterogeneous in space. 1984 data for. childhood leukaemia (ICD 204-8), from the Leukaemia Research Fund 1984 Data Collection Survey, show that rates within one year vary up to two-fold between different health regions (table). Although none of the observed/expected differences in the table are significant at the 5% level, the ratio for South West England is similar to that reported from West Berkshire. Other spatial data 1984 CHILDHOOD ACUTE LEUKAEMIA (AGE 0-9) IN SELECTED REGIONS OF 1
BRITAIN-r 1
I
incomplete to a degree ranging from 8% to 32%. The degree of completeness of other regional cancer registries is unknown. To make comparisons between the incidence of childhood leukaemia in a district with special data collection and national figures obtained routinely is likely to be misleading. In five districts in Wessex for which additional data are available supplement cancer registry data the average incidence of acute lymphoblastic leukaemia in children aged 0-9 years for the years 1972-82 ranged between 5-2and 6-4per 100000. The denominator used in the calculations was 1981 census data. A study of the geographical distribution of childhood leukaemia in Dorset shows a gradient of incidence from low rates in urban centres through higher rates in peripheral/commuter communities to higher rates still in rural areas, and a similar gradient from urban centres outwards has been noted elsewhere in Wessex (a detailed report is in preparation). This indicates that even where comprehensive data are available the balance between urban and rural communities must be taken into account when comparisons are being made. to
Central Administrative Offices, Royal Victoria Hospital, Bournemouth BH1 4HX
R. BARCLAY
SPLENECTOMY BEFORE BONE MARROW TRANSPLANTATION IN CHRONIC GRANULOCYTIC LEUKAEMIA
*Including smaller areas which are not detailed in table. tComputed on basis of 1981 census population data.
available in Yorkshire over a longer period of time suggest that variation is commonly seen in most leukaemia/lymphomas. When smaller areas are examined from the 1984 survey significant differences do emerge between expected and observed numbers, often for small numbers of cases in areas with tiny populations. Results highlighting specific areas are of interest but are usually uninterpretable because they are post hoc observations or contain unusual and biologically mixed subgroups of cases and do not account for the generality of cases found by a similar method of case ascertainment. This is one advantage of the 1984 Data Collection Survey which is currently being analysed with a view to understanding the true nature of the leukaemias and related conditions throughout certain parts of the UK. Uniform case ascertainment and analysis have been achieved without collection bias. Even so, cases very close together in space and time have been recorded from several localities in 1984 and 1985. Some clusters, after age and sex standardisation, are very unlikely to be due to chance and these are being fully investigated by the Leukaemia Research Fund cluster investigation group. The scientific community is not ignoring the investigation of such close case occurrences. Clusters need to be investigated responsibly, taking into account all known aetiological factors and not just those that reflect the prejudices of the media. Doctors who think that they have such case occurrence" worth investigating should contact me. I thank the many paediatric oncologists, haematologists, and histopathhave provided data for the Leukaemia Research Fund study.
ologists who
Leukaemia Research Fund 1984/85
Cookridge Hospital, Leeds LS16 6QB
Study, R. A. CARTWRIGHT
SiR,-We read with interest the paper by Dr Gratwohl and colleagues (Dec 7, p 1290) about the effect of splenectomy before bone-marrow transplantation (BMT) on survival in chronic granulocytic leukaemia (CGL). However, some of their conclusions ("debulking in this disease is of no value", "neither splenectomy nor splenic irradiation altered survival") seem premature and are not supported by statistical analysis. There are serious methodological objections to this multi centre non-randomised
study. Gratwohl et al compared a splenectomy group with a no splenectomy group. This comparison is not appropriate because splenectomised patients constitute a heterogeneous group (some with splenic irradiation and some without), and splenic irradiation may be more debulking than splenectomy. To evaluate the effect of splenectomy, it would have been better to have compared the splenectomy group with a no splenectomy/no irradiation group. Other comparisons to estimate the contribution to survival of an "asplenic status" (splenectomy plus splenic irradiation vs no splenectomy and no further therapy group) and of splenic irradiation (irradiation vs no irradiation/no splenectomy) would have been interesting. The second comparison Gratwohl et al did do-a simultaneous comparison of the three groups-does seem to have been done with heterogeneity in mind. If splenic irradiation may be more debulking than splenectomy, a test for trend’ would have been better, especially since the actuarial survival and the shape of the survival curves showed apparently ordered groups of risk (7lTo probability of survival in irradiated patients, 58% in splenectomised patients, and 49% in non-splenectomised, nonirradiated patients). These results must anyway be interpreted with caution because of the small numbers of patients at long-term risk in each category. The impact of splenectomy or splenic irradiation on survival in CGL patients treated by BMT must be investigated in prospective rar.domised trials before routine splenectomy is abandoned. Clinical
Haematology Section,
Haematology/Haemotherapy Service,
SIR,-Dr Barton and her colleagues report an average incidence of leukaemia in children between the ages of 0-9 years for the years 1972-1984 in West Berkshire of 5-88 per 100 000, compared with an average for England and Wales for the years 1972-81 of 4 - 3 per 100 000. In several districts in the Wessex region data about childhood leukaemia have been collected from a variety of sources, including the Wessex Cancer Registry, and paediatric, haematology and community child health departments, and these indicate that for acute lymphoblastic leukaemia the Wessex Cancer Registry is
"La Fe", Valencia, Spain
Hospital 46009
1 Peto R, Pike
requiring
MIGUEL A. SANZ GUILLERMO SANZ IGNACIO LORENZO
MC, Armitage P, et al. Design and analysis of randomized clinical trials prolonged observation of each patient. Br J Cancer 1977, 35: 1-39
SIR,-The report from the European Cooperative Group for Bone Marrow Transplantation provides important conclusions about the role of splenectomy in patients with CGL who are subsequently given a bone marrow transplant (BMT). Splenectomy was of no benefit in terms of survival, prevention of relapse, or incidence of