124

As in previous studies 5,9 some patients continue toproduce sputum which is positive on smear but nega-tive on culture, both during and after chemotherapy.Such results have been held to signify non-viablebacilli and to be of no clinical importance. This

finding in the last three months of chemotherapy wasnot taken to indicate a failure of chemotherapy : how-ever, 1 of the 12 patients in the six months groupwho had shown this feature did relapse. He hadalso produced a single positive culture in the last threemonths of chemotherapy.

Isolated positive cultures occurred after the cessa-tion of chemotherapy, as in other studies,5,9 but thefrequency of bacteriological examination in the pre-sent study has allowed differentiation from true

bacteriological relapse. In only 1 patient was chemo-therapy restarted by the physician on the basis of anisolated positive culture. The appearance of this

positive culture was not supported by any radiologicalor clinical deterioration and two subsequent sputumcultures produced before the reinstitution of chemo-therapy were negative: this patient has not been

regarded as having relapsed.The 5 patients who relapsed after stopping treat-

ment at six months did so with organisms sensitiveto streptomycin, isoniazid, rifampicin, and ethambu-tol, suggesting that the duration of chemotherapy hadbeen inadequate to kill off all sensitive bacilli.5 2 were

thought not to have taken their full course of chemo-therapy. All 5 have responded to reintroduction ofchemotherapy.

The Research Committee of the British Thoracic andTuberculosis Association acknowledges with gratitude the

cooperation of the many participating physicians and hospi-tal staff in England, Scotland, Wales, and Ireland; CIBALaboratories and Lepetit Pharmaceuticals for financial sup-port ; and Mrs C. P. Davison, Mrs P. Hindshaw, Mrs F.

Hoare, and Mrs L. Hunt for their assistance.

Requests for reprints should be addressed to Dr I. A.

Campbell, Department of Respiratory Diseases, City Hospital,Edinburgh EH10 5SB.

REFERENCES

1. Ross, J. D., Horne, N. W., Grant, I. W. B., Crofton, J. W. Br.med. J. 1958, i, 237.

2. Ross, J. D. Bull. Un. int. Tuberc. 1959, 29, 710.3. Crofton, J. W. Br. med. J. 1960, ii, 679.4. Pearce, S. J., Horne, N. W. Lancet, 1974, ii, 641.5. Report to the Medical Research Council by their Tuberculosis

Chemotherapy Trials Committee. Tubercle, 1962, 43, 201.6. Pande, B. R., Martischnig, K. M., Feinmann, L. ibid. 1970, 51, 39.7. Lal, S. ibid. 1969, 50, 269.8. East African/British Medical Research Councils. Lancet, 1972, i,

1079.

9. East African/British Medical Research Councils. ibid. 1973, i, 1331.10. East African/British Medical Research Councils. ibid. 1974, ii, 237.11. Marks, J. Publ. Hlth Serv. Monogr. Ser. no. 5; laboratory methods,

I, 14. H.M. Stationery Office, 1974.12. Leat, J. L., Marks, J. Tubercle, 1970, 51, 68.13. Tuberculosis Chemotherapy Centre, Madras. Bull. Wld Hlth Org.

1960, 23, 535.14. Simon, G. Br. med. J. 1966, ii, 491.15. Brouet, G., Roussel, G. in XVII Congres national de la tuberculose

et des maladies respiratoires; p. 27. Paris 1974.16. East African/British Medical Research Councils. Lancet, 1974, ii,

1100.

17. Poppe de Figueiredo, F., Brito, A. A., Valle, J. H. L., Tavares, P. M.,Trannin, P. L. Proc. XXIInd Tuberc. Conf. 1973. (In the press.)

18. Lal, S., Singhal, S. N., Burley, D. M., Crossley, G. Br. med. J.1972, i, 148.

19. Baron, D. N., Bell, J. L. Tubercle, 1974, 55, 115.

LEUKÆMIA AND LYMPHOMA PATIENTSINTERLINKED BY PRIOR SOCIAL

CONTACT *

STEPHEN C. SCHIMPFF CAROL R. SCHIMPFFDORAH M. BRAGER PETER H. WIERNIK

Section of Medical Oncology, Baltimore Cancer ResearchCenter, National Cancer Institute, University of Maryland

Hospital, Baltimore, Maryland 21201, U.S.A.

Summary Patients with either leukæmia or

lymphoma were asked if they had closepersonal associations with other patients before theonset of disease. Initial interviews indicated thatseveral patients could be interlinked into social clusters.Tumour-registry records were used to contact eachpatient (or a surviving relative) diagnosed during theyears 1964-73 in three areas of West Virginia. Closepersonal associations, antedating the onset of diseasein 1 or both individuals of each linkage pair, weredetected in 14 of 23 (61%), 14 of 22 (68%), and 6of 8 (75%) patients from these three areas during thisten-year period. In addition, among 28 randomlyselected patients with Hodgkin’s disease from variousparts of the United States, 10 (36%) had direct orindirect close personal associations with 17 other

patients with leukæmia or lymphoma. Patients withleukæmia or lymphoma frequently are interlinked byprior close personal associations to other patients withthese diseases.

Introduction

THERE is increasing interest in the possibility of

person-to-person transmission of leukarmias or

lymphomas in a manner similar to that shown forfowl and cats.1-3 Most human studies have concen-trated on one disease entity such as Hodgkin’s diseaseor acute lymphocytic leukaemia.H2 For example,Vianna’s group 4,5 has linked together 31 patients withHodgkin’s disease on the basis of prior personalassociations, and Kemmoona has found 5 patientswith acute lymphocytic leukaemia and another withlymphosarcoma who could be interlinked through theirsocial contacts.Our investigations of Hodgkin’s disease in one area

and leukaemia in another resulted in the detection of

large numbers of socially interlinked patients. Inthese and other areas, however, multiple social linkageswere found among patients with all forms of leuk-xmia and lymphoma rather than with only a singletumour type. These findings indicate that manyindividuals with either leuksemia or lymphoma canbe interlinked into social clusters based on priorpersonal associations.

Methods

Three separate areas of West Virginia were studied.In each area, patients with leukaemia or lymphoma fromthe Baltimore Cancer Research Center had reported know-ledge of one or more close personal friends with leukaemiaor lymphoma living near their homes. The West Virginia

* Based on a paper presented at the Fourteenth InterscienceConference on Antimicrobial Agents and Chemotherapy,held in San Francisco in September, 1974.

125

tumour registry, hospital records, and/or reports from

personal physicians were used to confirm or reject the

reported diagnosis. If the diagnosis was confirmed, thefriend was asked if he and the index patient had indeedhad a close personal relationship before the onset of dis-ease in either himself or the index patient. He was alsoasked if he was personally familiar with other individualswith these diseases, and these cases were confirmed andinterviewed in the same way until all leads were exhausted.

Close personal associations were defined as householdcontacts (spouses, sibs, parent/child, or other pairs of indi-viduals residing continuously or intermittently in the sameprivate home), social friendships (dating partners and otherclose associations usually but not necessarily involving fre-quent visits to each other’s homes), or other associationswhich dictated close social interaction (e.g., 2 telephonelinemen working from the same truck). 2 students in thesame class who were not personal friends, or 2 men work-ing in different sections of the same large factory, wereclassified as acquaintances and not as close personalfriends. Whenever possible, both parties to each associa-tion were contacted to ensure concordance. Only asso-ciations which had existed before the onset of diseasein one of the two patients were used for this study, andrelatiqnships which developed after tumour diagnosis inboth individuals (e.g., in the clinic) were ignored. If a

patient was dead, a surviving relative or other knowledge-able person, if available, was contacted to verify associa-tions.Once the case-to-case interviews were completed it was

noted that most interlinked cases occurred within reason-ably close geographic proximity to each other. There-fore, the three study areas were further defined on the

basis of political or census boundaries. Using the WestVirginia tumour registry we obtained a complete list ofall cases of leukaemia diagnosed during the ten years 1964-73. Patients diagnosed while residing in the three definedareas who were previously unknown to us were contactedto determine if they knew other patients in their area.

2 patients were not contacted (at the request of their per-sonal physicians), and some dead patients had no suitablesurviving relative to interview.

Results

In each figure, the patients with leuksemia or

lymphoma are represented by boxes which containinitials (fictitious to preserve anonymity), diagnosis,year of diagnosis, and year of death.

Close personal associations between patients are

represented by interconnecting solid lines. Inter-

mediary contacts are represented by circles withinitials designed to identify their association with apatient-e.g., F. T. in fig. 1 is the father (F) of thepatient J. T. The nature of each relationship (e.g.,parent/child or social friends) is indicated along thesolid line. Some boxes (patients) or circles (inter-mediaries) have adjoining small arrows with initials.These represent close personal associations with otherpatients who live outside the area defined by the

political or census boundaries. Finally, the figuresare arranged chronologically so that patients diagnosedin 1964 or earlier are at the top and those diagnosed in1973 or later are at the bottom.

Fig. 1-Area 1: clustering of leukaemia and lymphomas.Patients represented by squares which include initials, diagnosis, year of diagnosis, and year of death. Intermediary contacts

represented by circles with initials. Social links are represented with solid lines. Index patients are indicated by large blackarrows. The figure is arranged chronologically with the earliest diagnosed patients at the top. Links to patients residing outsidethe area are illustrated with small arrows.

A.L. =acute leukaemia. H.D. =Hodgkin’s disease.A.M.L. =acute myelocytic leukaemia. LYM==lymphoma.A.L.L. =acute lymphocytic leukaemia. LSA=lymphosarcoma.c.M.L. =chronic myelocytic leukaemia. R.C.S. =reticulum-cell sarcoma.C.L.L. =chronic lymphocytic leukaemia. M.F. =myelofibrosis.EOSL = eosinophilic leukaemia.

126

Area 1 (fig. 1)This is a rural coalmining county with a stable

population of about 5000. Two index patients(B. N. and J. T., large arrows) had A.M.L., and theyare linked via the father of one of them who worked

closely with the other index patient in the same smallcoalmine. B. N. is linked directly to B. D. (acute1eukaemia). Another close friend of B. N. is A. Q.who developed lymphosarcoma in 1968 and A. C.who was found to have lymphosarcoma in 1970. BothB. N. and F. T. worked in the same mine as thefather of a young girl (D. D.) who developed A.L.L.in 1969. J. T. (the other index patient) had directclose associations with H. H., S. T., and C. M.

Multiple other close personal associations either

directly (patient to patient) or indirectly (patient tointermediary to patient) are illustrated. In total, 18patients are interlinked into this one cluster. 2 otherpatients (D. E. and K. U.) were indirectly linked butthey denied any close personal associations with

patients in the cluster.27 cases are illustrated in fig. 1 of which 23 are

all the cases reported to the tumour registry for theyears 1964-73. 23 is the approximate number expectedfor a population of 5000 in West Virginia. 14 of the23 (61%) were socially interlinked; 2 were linked toeach other but to no-one else; and 7 could not beinterlinked with any other patient.

Area 2 (fig. 2)This area covers three small communities with both

industrial and rural elements (total population about5000). A mountain ridge divides the area into twosections (dotted line in fig. 2) that are distinct andseparate except that they share a common high schoolwhere many residents of the two sections first meet.There are many patient-to-patient and patient-to-

intermediary-to-patient associations. Notable examplesinclude a subgroup with Hodgkin’s disease andanother cluster of patients with A.M.L. There are 8

patients with Hodgkin’s disease; 4 (E. G., E. D., W. I.,and D. T.) are closely and directly interlinked throughtheir prior personal associations. Of the 5 patientswith A.M.L., 4 can be interlinked: F. G. died of A.M.L.in 1952; his sister married W. C., and this man (whonever met F. G.) developed A.M.L. in 1966. Both F. G.and his sister were close friends of L. I. (A.M.L. in 1968)and E. N., a close friend of L. I., was found to haveA.M.L. in 1971.This group of 4 patients with A.M.L. are interlinked

to 2 patients with C.M.L. (L. N. and D. T.), 1 patientwith either A. or C.M.L. (G. E.), and 1 patient withmyelofibrosis (M. C.).Of additional interest is a case of lymphosarcoma

(H. X.) in 1965; C.M.L. was diagnosed in this man’swife only five years later.We found close personal associations interlinking

19 individuals with leuksemia and lymphoma fromarea 2. For 1964-73 22 cases were reported to thetumour registry which approximates the expectednumber of patients. 14 of these 22 (68%) are inter-linked into the social cluster shown in fig. 2.

Area 3 (fig. 3)This is a suburban-rural area on the outskirts of a

major city. Most people have lived in the generalvicinity throughout their lives.The index patient (B. I.) reported that several people

who were living or had lived along the one major roadand adjoining cul-de-sacs in this area had haematologicmalignancies. The nature of the area is such that mostindividuals are at least acquaintances with each other;however, among the patients with leukaemia or

lymphoma we were able to interlink 14 patients on the

Fig. 2-Area 2: clustering of teukaemias and lymphomas.

127

basis of close personal associations. In fig. 3 patientsresiding in the area at the time of diagnosis are withinthe large dotted line rectangle. Those shown outsideof the rectangle are previous residents or individualswho work but do not live in this area. The arrowsindicate linkages to patients who neither live nor workin this census defined area.Pertinent examples include the development of

lymphosarcoma in 1968 in J. U., who lived above thegeneral store owned by the father of B. I., and thedevelopment of lymphosarcoma in 1973 by C. N. whohad lived next door to a young boy with A.L.L. (A. T.).Perhaps the most interesting observation was that

two years after this study began a close friend andneighbour of B. I. developed Hodgkin’s disease

(M. E.). This woman had acted as a housekeeperintermittently over a five-month period in 1972 whileB. I. was in hospital.During the ten years 1964-73 there were 8 newly

diagnosed patients residing in the area who were

reported to the tumour registry. 6 of the 8 (75 %) canbe socially interlinked.

Areas 1-3

During the ten-year period 1964-73 53 patients withleukaemia and lymphoma were reported to the tumourregistry. 34 (64%) could be interlinked to 1 or moreother patients in their own areas through close per-sonal associations.

Random Search for Other LinkagesThese three areas (plus two others under investiga-

tion) have multiple patient linkages, which suggeststhat social contact before disease onset is common

among patients with leukaemia and lymphoma. How-ever, these areas might not be typical since

RANDOMLY SELECTED CLINIC PATIENTS WITH HODGKIN’S DISEASE:ASSOCIATIONS WITH OTHER PATIENTS WITH LEUKAEMIA OR

LYMPHOMA

* Patient-to-patient close personal contact.t Patient-to-intermediary-to-patient contact.H.D. = Hodgkin’s disease. A.L. = acute leukaemia. C.M.L. = chronic

myelocytic leukaemia.

investigations began after a patient had reportedpersonal contact with other patients. We decided totest the hypothesis that close personal associationsare common among these patients and not as an

occasional chance occurrence.

28 patients with Hodgkin’s disease attending ourclinic were selected at random for interviews. Patients

residing in any of the areas under investigation wereexcluded. 10’of the 28 (36%) patients indicated thatthey had one or more direct or indirect close personalassociations with other patients who had Hodgkin’sdisease or leukxmia (table). 14 of the patients livedin areas of high population mobility; only 1 of these

patients could recall a close association with anotherpatient. However, 9 of the 14 (64%) patients fromareas of low population mobility did report one ormore direct or indirect close personal contacts who hadeither leuksemia or lymphoma.

Fig. 3-Area 3: clustering of leucaemias and lymphomas.

Patients within the large rectangle resided in area at time of diagnosis; those outside the rectangle either previously lived in areaor now work in the area.

128

The 10 patients who responded affirmatively (table)reported direct patient-to-patient contact with 9individuals with Hodgkin’s disease, 1 with acute

leukxmia, and 1 with C.M.L. In addition there were6 patient-to-intermediary-to-patient associations

(Hodgkin’s disease 4, acute leukxmia 2).These observations among randomly selected

patients with Hodgkin’s disease from low-mobilityregions confirm that the associations found amongpatients in the three West Virginia areas were neitheran unusual nor a chance finding.

Discussion

Earlier investigations into the epidemiology ofhuman leukxmias and lymphomas have generally beendirected toward one specific histologic type oftumour.-12 Very little success has been obtained in thesearch for time-space clusters.13,14 The earliest detailedinvestigation suggesting that A.L.L. might be due toan infectious agent was done by Heath and Hasterlik 15They found that 7 of 8 cases of childhood leukaemia ina four-year period in Niles, Illinois (representing afourfold increase in the expected incidence), wereassociated with children attending one school and wereaccompanied by the parallel appearance of a" rheumatic-like " illness among other children in thesame school. Although there was no evidence ofdirect interpersonal contact, the social situation of thechildren was such that Heath and Hasterlik suggestedthat an agent with high infectivity but low patho-genicity might have been the caused

Since that time others have reported groups ofpatients with either leukaemia or lymphoma inter-related by geography or social contacts 4-12 Ingeneral, spatial clustering has been confined to smallgroups of patients with the same disease often inrelatively isolated areas. If transmissible agents with

long incubation periods either cause or allow expres-sion of leuksemia or lymphoma, it is not surprising thatgeographic clustering based on the time of diagnosis isdifficult to establish. More pertinent would be to

establish the possible cluster based on exposure whichmight have happened months or years before. This is,in effect, what our study and that of Vianna et allhave attempted to do by determining past personalassociations.

Vianna’s group 4,5 described patients with Hodgkin’sdisease in Albany, N.Y., who were interlinked sociallyeither on a case-to-case or a case-to-contact-to-casebasis. There was no apparent geographical clustering-all linkages were based on prior personal associa-tions. The same group of investigators have reportedthat high-school students in a metropolitan area whohad a teacher with Hodgkin’s disease had a higherincidence of Hodgkin’s disease in the ensuing five toten years than did students in similar schools whereno teacher had Hodgkin’s disease .16

We found, unexpectedly, that the individuals initiallyinterviewed reported close associations not only withother patients who had the same diagnosis but alsowith patients who had any type of leukaemia or

lymphoma. We therefore continued the investigation

by asking each patient about direct or indirect con-tacts with patients who had either leukaemia or

lymphoma.In this regard, it is interesting that the Gross

leukaemia virus can produce, in the same inbred

colony of mice, diseases ranging from A.L.L. to histio-cytic lymphoma,17 Horizontal transmission of thefeline-leukxmia virus with a long-incubation periodhas been described in the domestic cat.2,3 Diseasedanimals shed the virus via saliva and urine. About athird of normal cats residing in the same householdtend to become infected with the transmitted viruswhich is detectable in platelets and leucocytes.Approximately a quarter (888 times the expectedincidence) of these infected cats developed lympho-sarcoma, refractory anaemia, or infectious peritonitisover the six months after virus detection.2,3 Theseobservations that single virus strains can produce aspectrum of diseases following a prolonged incubationperiod support the possibility that something similarmight happen in man.Our data, Vianna’s investigations,’,-1,16 and the other

reports of similar although smaller groups linked byprior social contacts s-12 strongly indicate that priorpersonal contact is coinmonplace among these patients.Our own study emphasises that patients with any typeof leukaemia or lymphoma tend to fit into socialclusters.

Interviews with 28 randomly selected patients withHodgkin’s disease revealed that 10 had direct (9)and/or indirect (4) contact with other patients whohad leukaemia or lymphoma. Heath et al.,t8 report-ing on interviews with 36 patients with Hodgkin’sdisease residing in Atlanta, Georgia, found only 2 whoindicated personal knowledge of other patients withthe same disease. These findings are similar to ourpatients with Hodgkin’s disease selected from high-mobility urban areas. However, 64% of our patientsfrom low-mobility areas responded that they did havedirect or indirect close associations with other patientswith Hodgkin’s disease or leukaemia.

It seems clear that patients with leukxmia or

lymphoma, especially those from low populationmobility areas, are often linked together by close per-sonal contacts before the onset of disease in one orboth individuals in each linkage pair. Pike andSmith 19,20 suggest that individuals living in low-

mobility areas may tend, as a general rule, to know oneanother. While this is very likely to be true, the

implication of our investigation and that of others isthat the relationships reported have been ones of aclose personal nature and not mere local acquaintance-ship. In this regard, however, the very closeness ofmany of the associations reported suggests that theseare not just average casual acquaintances. For

example, in area 1, B. N. and B. D. were lovers, J. T.and H. H. rode to church together for years, and J. T.and S. T. were cousins and very close friends. In area 2,H. X. and L. X were married, F. G.’s sister marriedW. C., B. L. and W. T. had dinner and played bridgetogether regularly, and M. C. and J. D. were next-doorneighbours. In area 3, J. H. and J. T. were next-doorneighbours as were A. T. and C. N., M. E. was a house-

129

keeper for B. 1., and H. C. was a maid for D. L., whoseclose friend was A. T. Among the randomly selectedclinic patients with Hodgkin’s disease, 1 girl becameill within a year of steadily dating a man with

Hodgkin’s disease, a girl developed Hodgkin’s diseasewithin two years of the same diagnosis in her closest

! friend, acute leukaemia developed in a child of a familywho transiently housed a patient with Hodgkin’sdisease, and Hodgkin’s disease occurred in a youngman who helped a friend with Hodgkin’s diseaserebuild his house.

These instances of close personal contact betweenindividuals with leukaemia or lymphoma cannot belightly dismissed. Nevertheless, there is still need forproof that the associations reported among patients aremore common than would be expected for individualsliving in the area under investigation. We are

attempting to resolve this issue in a fourth area. Agroup of individuals matched for age, sex, race, andhome location has been selected to determine if theywill or will not have the same frequency of past closesocial interactions as did the patient group.The data now available indicate that patients with

either leukaemia or lymphoma often have had direct orindirect close personal associations with other similarpatients. This finding is consistent with the hypothesisthat some forms of human leukaemia and lymphomamay be associated with the transmission of some

agent(s) from person to person.

Major and continuing assistance was offered by Mrs FrancesShaver, Dr William L. Cook, and their associates at the WestVirginia Division of Cancer Control tumour registry and bythe many patients, their friends, and families who kindlyallowed an intrusion into their personal lives. The commentsof Dr George Comstock and Dr J. N. P. Davies are gratefullyacknowledged.

Requests for reprints should be addressed to S. C. S.

REFERENCES

1. Rubin, H., Cornelius, A., Fanshier, L. Proc. natn. Acad. Sci.U.S.A. 1961, 47, 1058.

2. Hardy, W. D., Hess, P. W., Essex, M., Cotter, S. Nature, 1973,244, 266.

3. Hardy, W. D., Geering, G., Old, L. S., deHarven, E., Brodey, R. S.,McDonough, S. Science, 1969, 166, 1019.

4. Vianna, N. J., Greenwald, P., Davies, J. N. P. Lancet, 1971, i, 1209.5. Vianna, N. J., Greenwald, P., Brody, J., Polan, A. K., Dwark, A.,

Mauro, J., Davies, J. N. P. Ann. intern. Med. 1972, 77, 169.6. Kemmoona, I. Lancet, 1974, i, 944.7. Reske-Nielson, E., Peterson, J. H., Sogaard, H., Jensen, K. B.

ibid. p. 210.8. Klinger, R. S., Minton, J. P. ibid. 1973, i, 168.9. Heath, C. W., Everett, J. R., Steward, J. T., Daines, J., Daines,

P. H. ibid. p. 669.10. Levine, P. H., Sandler, S. G., Komp, D. M., O’Conor, G. T.,

O’Conor, D. M. New Engl. J. Med. 1973, 288, 562.11. Gilmore, H. R., Zelesnick, G. Penn. med. J. 1962, 65, 1047.12. Parker, J. E. Lancet, 1974, i, 210.13. Merrington, M., Spicer, C. C. Br. J. prev. soc. Med. 1969, 23, 124.14. Adlerson, M. R., Nayak, R. ibid. 1971, 25, 168.15. Heath, C. W., Hasterlik, R. S. Am. J. Med. 1963, 34, 796.

16. Vianna, N. J., Polan, A. K. New Engl. J. Med. 1973, 289, 499.17. Gross, L. Oncogenic Viruses. New York, 1970.18. Heath, C. W., Rosenstock, J. G., Lobdell, G. Lancet, 1971, ii, 426.19. Smith, P. G., Pike, M. C., Kinlen, L. S. ibid. 1973, i, 433.20. Pike, M. C., Smith, P. G. Cancer, 1974, 34, 1390.

TREATMENT OF SMALL-CELL CARCINOMA

OF BRONCHUS

A. H. LAING R. J. BERRY*

Radiotherapy Department, Churchill Hospital,Headington, Oxford OX3 7LJ

C. R. NEWMAN

Horton General Hospital, Banbury, Oxon

P. SMITH

D.H.S.S. Cancer Epidemiology Clinical Trials Unit,Oxford

Summary A randomised trial comparing radio-therapy with a multiple chemotherapy

regimen in 68 patients with small-cell carcinoma ofbronchus is reported. Although overall survival waspoor, radiotherapy resulted in significantly longersurvival, better amelioration of symptoms, and lessside-effects than chemotherapy. Patients achievingcomplete or partial remission on radiotherapy relapsedmainly with extrathoracic disease, in contradistinctionto those treated with chemotherapy whose thoracicdisease recurred.

Introduction

A RANDOMISED trial of the treatment of inoperablecarcinoma of the bronchus was started in the Oxfordregion in 1970 as a second stage to the trial reportedby Durrant et al.,1 patients being entered into the trialfrom seven participating centres in the region. All

patients presenting with carcinoma of the bronchuswere eligible for entry into the trial except those

patients so ill that symptomatic treatment only wasindicated. This report considers one group of thetrial patients, consisting of all those patients initiallydiagnosed as having small-cell carcinoma of thebronchus (W.H.O. classification). These tumoursare known to have a different morphological charac-teristic, natural history, prognosis, and response to

treatment to other histological types,2-’ and they havetherefore been considered separately.

Patients were assigned at first presentation to receivetreatment with either radiotherapy or intermittentcombination chemotherapy.

TreatmentRadiotherapy

Treatment was by supervoltage irradiation usingparallel-opposed fields to the whole mediastinum. Thesupraclavicular region was treated in all patients by anextension to both anterior and posterior fields if glandswere present, or otherwise by an extension to the anteriorfield only. The dose employed was 3500 rads midlinein four weeks treating daily (3500 rads in 20 fractionsin 25 days), or 3000 rads treating three times weekly(3000 rads in 12 fractions in 25 days). If residualtumour was present, a further 1000 rads in 1 week couldbe given (1000 rads in 5 fractions in 4 days).

ChemotherapyTreatment was by pulsed administration of a com-

bination of nitrogen mustard (mustine hydrochloride),

* Present address: M.R.C. Radiobiology Unit, Harwell, Didcot, Oxon.