1116
Prompted by reports that reovirus-like particles are commonly associated with acute childhood gastroenteritis,6-’o ’6 we searched for evidence of infection by this agent in children not infected with either classic bacterial pathogens or L.T.-positive E. coli. An antigenic relation has been demonstrated between the reoviruslike agent found in stools of children with diarrhoea and a morphologically similar virus, N.C.D.V.24-:U. Immunofluorescent-staining, using primary B.E.K. cell monolayers infected with N.C.D.V., provides a sensitive method for detecting serological responses to infection with the human reovirus-like agent. 18-19 Paired sera from 31 children were tested for immunofluorescent-staining antibodies to N.C.D.v. 35% of children tested showed evidence for a recent infection, a percentage similar to that reported in other studies of infection by the reovirus-like 8 10 19 29 agent in children in hospital with enteritis.7 Our seroconversion data do not prove that the cause of diarrhoea in these children was the reovirus-like agent, since we were unable to perform electron-microscopic studies that would have detected the specific agent in their stools. However, the increase in antibody that coincided with their admission to hospital with enteritis provides strong evidence that most, if not all, of the 11 children had enteritis due to infection with the reoviruslike agent. Indeed, because of the reported lack of complete concordance between serological assays with the human and N.C.D.v. antigens,’9 our results may have underestimated the true frequency of infections caused by the reovirus-like agent in the study group. 9 of the 11 children were admitted to hospital during winter months, a time of year when infections by the reoviruslike agent have been reported to be most common.8 10 29 The results of our one-year surveillance of children admitted to hospital in Boston because of diarrhoea further emphasize the aetiological involvement of the reovirus-like agent. They also suggest that L.T.-positive E. coli may not account for a significant percentage of the previously undiagnosed, endemic childhood cases of diarrhoea in such a setting. Finally, the failure to identify a causal agent in two-thirds of these children indicates the need for continued study of the infectious causes of childhood diarrhoea.
ENTEROTOXIGENIC ESCHERICHIA COLI AND IDIOPATHIC DIARRHŒA IN BANGLADESH D. R. NALIN M. RAHAMAN
Requests for reprints should be addressed to D. S., Children’s Hospital Medical Centre, 300 Longwood Avenue, Boston, Massachusetts 02115, U.S.A. REFERENCES
Melnick, J. L., Blattner, R. J., Stephenson, W. B., Robinson, N. M., Burkhardt, M. A. Am. J. Epidem. 1970, 92, 33. Cramblett, H. G., Siewers, C. M. F. Pediatrics, Springfield, 1965, 35, 885 Gorbach, S. L., Khurana, C. M. New Engl.J. Med. 1972, 287, 791. Sack, R. B., Hirschhorn, N., Brownlee, I., Cash, R. A., Woodward, W. E., Sack, D. A. ibid. 1975, 292, 1041. Rudoy, R. C., Nelson, J. D. Am. J. Dis. Child. 1975, 129, 668. Bishop, R. F., Davidson, G. P., Holmes, I. H., Ruck, B. J. Lancet, 1974, i,
1. Yow, M. D., 2. 3. 4.
5. 6.
149. 7. Flewett, T. H., Davies, H., 1974, 27, 608.
Bryden,
A.
S., Robertson, M. J. J clin. Path.
M. YUNUS G. CURLIN
Cholera Research Laboratory, Johns Hopkins University, International Center for Medical Research, G.P.O. Box no.
128, Dacca, Bangladesh
Fæcal Escherichia coli from Bangalee patients with idiopathic diarrhœa were tested in the Chinese hamster ovary cell (C.H.O.) assay to detect production of thermolabile enterotoxin (L.T.). C.H.O.-positive E. coli produce both L.T. and a thermostable toxin (S.T.). C.H.O.-positive strains were found in 19.2% of all cases and in 70% of the most severely ill patients with non-vibrio cholera. E. coli which produce S.T. alone were identified in a third of the C.H.O.-negative cases. Enterotoxigenic E. coli were found in 55% of inpatients with idiopathic diarrhœa and, if an ætiological role is confirmed, may be one of the commonest causes of tropical diarrhœa.
Summary
Introduction
HUMANS and animals with idiopathic diarrhoea may harbour faecal Escherichia coli which produce both thermostable
8. 9.
10. 11. 12.
13.
14.
We thank the nursing staff and the house staff on division 27 at the C.H.M.C. for their help and cooperation; private physicians who collected convalescent-phase specimens for us; and Dr M. Oxman, Dr T. Thornhill, and Dr A. Smith for their helpful suggestions. L.T.-producing strains of E. coli were supplied by Dr R. B. Hornick (B7A, 334A, and B2C) and Dr R. B. Sack (Nagle 6). Adrenal cells used in the enterotoxin assay were supplied by Dr S. T. Donta and Dr R. B. Sack. This study was supported by a National Institute of Allergy and Infectious Diseases training grant and by the U.S. Army Medical Research and Development Command.
J. C. MCLAUGHLIN
15. 16.
(s.T.) and thermolabile (L.T.) enterotoxins
Middleton, P J., Syzmanski, M. T., Abbott, G. D., Bortolussi, R., Hamilton, J. R. Lancet, 1974, i, 1241. Kapikian, A. Z., Kim, H. W., Wyatt, R. G., Rodriguez, W. J., Ross, S, Cline, W. L., Parrott, R. H., Chanock, R. M. Science, 1974, 185, 1049 Davidson, G. P., Bishop, R. F., Townley, R. R. W., Holmes, I H., Ruck, B. J. Lancet, 1975, i, 242. Sack, R. B., Gorbach, S. L., Banwell, J. G., Jacobs, B , Chatterjee, B. D., Mitra, R. C.J. infect. Dis. 1971,123, 378. Gorbach, S. L., Banwell, J. G., Chatterjee, B. D., Jacobs, B, Sack, R B J. clin. Invest 1971, 50, 881. Dupont, H. L., Formal, S. B., Hornick, R. B., Snvder, M J, Libonati, I P., Sheahan, D. B., LaBrec, E. H., Kalas, J. P New Engl.J Med 1971, 285, 1. Dean, A. G., Ching, Y. C., Williams, R. G., Harden, I.. B. J. infect Dis 1972, 125, 407. Punyashthiti, K., Finkelstein, R. Infect. Immun. 1971, 4, 473. Bishop, R. F., Davidson, G. P., Holmes, I. H., Ruck, B J. Lancet, 1973, ii, 1281.
Flewett, T. H., Bryden, A. S., Davies, H., Woode, G. N., Bridger, J. C. Derrick, J. M. ibid. 1974, ii, 61. 18. Flewett, T. H., Bryden, A. S., Davies, H., Morris, C. A. ibid. 1975, i, 4 19. Kapikian, A. Z., Cline, W. L., Mebus, C. A., Wyatt, R. G, Kalica, A R, James, H. D., VanKirk, D., Chanock, R. M., Kim, H. W. ibid p 1056 20. Edwards, P. R., Ewing, W. H. Identification of Enterobacteriaceæ. Minnea17
polis, 1972. 21. Donta, S. T., Moon, H. W., Whipp, S. C. Science, 1974,183, 334 22. Sack, D. A., Sack, R. B. Infect. Immun. 1975,11, 334 23. Donta, S. T., Sack, D. A., Wallace, R. B., DuPont, H I.., Sack, R B New Engl. J. Med. 1974, 291, 117. 24. Fernelius, A. L., Ritchie, A. E., Classick, L. G., Norman, J O, Mebus, C A. Arch. Virusforsch. 1972, 37, 114. 25. Mebus, C. A., Kono, M., Underdahl, N. R., Twiehaus, M J Can. vet J 1971, 12, 69. 26. Welch, A. B., Twiehaus, M. J. Can. J. comp. Med 1973, 37, 287 27. Dolin, R., Blacklow, N. R., Wyatt, R. G., Sereno, M. M Infect Immun
1972, 6, 958. 28. Mebus, C. A., Underdahl, N. R., Rhodes, M. B., Twiehaus, M. I Res Bull Neb. agric. Exp. Stn. 1969, 233. 29. Bryden, A. S., Davies, H. A., Hadley, R. E, Flewett, T. H, Morris, C A. Oliver, P Lancet, 1975, ii, 241. 30. Chrystie, I. L., Totterdell, B., Baker, M. J., Scopes, J. W., Banatvala, J F ibid. p. 79. 31. Smith, H. W., Gyles, C. L.J. med Microbiol. 1970, 3, 387 32. Evans, D. G , Evans, D. J., Pierce, N. F. Infect. Immun 1973, 7, 873 33. Donta, S. T. Unpublished 34. Sack, R. B., Jacobs, B., Mitra, R.J. infect. Dis. 1974, 129, 330
1117
(S.T.+/L.T.+).1-3 Some animal strains produce S.T. alone (S.T.+/0)3 but human S.T.+/O strains have been reported from only 6 patients.24 S.T.+/L.T.+ strains be identified in the Chinese hamster ovary (c.H.o.) assay5 due to morphological changes induced in c.H.o. cells by L.T. There is no simple s.T. test. The mouseloop test has been used,6 but S.T. may decrease intestinalloop absorption without causing actual fluid accumulation,1-7 so false negatives are best avoided by measuring post-toxin intestinal net water fluxes. In this study the prevalence of S.T.+/O and S.T.+/L.T.+ strains in rectal swab cultures of Bangladesh patients with idiopathic diarrhoea was determined by measuring c.H.o. cell and dog-intestinal water flux responses to E. coli culture can
supernatants. Patients and Methods Between Nov. 10, 1974, and July 8, 1975, all verified cases of idiopathic diarhoea among 92 838 participants in a cholera toxoid vaccine trial were reviewed. Birth dates of the majority were known. All females over 1 year old and all males 1-14 years old were included. Most adult males and infants under 1 were excluded due to trial design. A single diarrhoea hospital serves the censused village trial area (population 167 610). Fæcal E. coli from 96% of trial participant patients whose cultures were negative for Vibrio, Shigella, and Salmonella species (123 inpatients and 370 outpatients) and 52 other idiopathic diarrhoea patients’ E. coli strains were tested for S.T.+/L.T.+ in the c.H.o. assay.s These were classified as C.H.O.+ (= S.T.+/L.T.+) or c.H.o.-. A random group of 20% of c.H.o.- inpatients’ E. coli culture supernates and 24 others (including C.H.O.+ controls) were tested for S.T.+/0 in dogs. Patients’ pre-treatment rectal swabs cultured on MacConkey’s, S. S., and Monsur’s agar showed E. coli but no known pathogens. Patients received intravenous and oral replacement of diarrhœal fluid and electrolyte losses. After cultures were plated, almost all patients received tetracycline by mouth 6-hourly for 5 days (250 mg for adults and 125 mg for children). Ten E. coli colonies from each of 545 patients’ MacConkey’s plates were pooled and stored on blood-agar base slants at room temperature in the dark. E. coli pools were confirmed biochemically. For c.H.o. assay trypticase soy broth with 0-6% yeast extract and 0-2% glucose at pH 8 was inoculated with E. coli and incubated 48 hours at 35 °C without shaking. Supernates were tested after cold centrifu-
gation. supernates (and c.H.o.+ were Evans’ controls) prepared by method8 at pH 8. Heated and unheated 60-100 x concentrated hypotonic (4-50 mosmol/1) supernates were tested for s.T. in adajcent 17 cm dog jejunal loops.1 Heated supernates were placed in a boiling-water bath for 20 minutes and cooled before testing. Each loop received 4.0 ml of concentrate. For
dog-loop
tests
C.H.O.
TABLE I-MEAN CLINICAL DATA ON
Twenty of the pooled isolates tested for S.T.+/L.T.+ and S.T.+/0 were from severely dehydrated adult inpatients at Dacca (chiefly males) with typical non-vibrio cholera.9 These patients had a preadmission duration of less than 1 day, were typically admitted in shock, and had total diarrhoea volumes equal to at least 5% of body-weight. They received intravenous therapy but no antibiotics. Results C.H.O. Assay and Dog-loop Tests Most patients presented with mild watery diarrhoea and vomiting of short duration, often with mild abdominal pain but usually without mucoid or bloody stool. 33% of inpatients’ and 15% of outpatients’ E. coli isolates (19-2% of all cases) were C.H.O.+ Seven of twenty-three randomly selected c.H.o.inpatients’ isolates produced only S.T. S.T. was detected by exposure of 28 dog-jejunal loops to six heated and twenty-two unheated aliquots of supernates from nine strains. 20 minutes later, loop absorption (0.11±0.05 ml/min/loop) changed to secretion (+0-05+0-05 ml/
min/loop;
mean
±S.D., P≤0.001). Absorption
re-
mained normal up to 6 hours after rinsing filtrates out. This response is typical of S.T’! Five control loops with no supernate and twenty-two with non-toxic supernates absorbed -0-08 to -0-10 ml/min/loop for up to 6 hours. In seventeen positive control loops S.T.+/L.T.+ supernates induced a change from absorption (-0.13 +0-05) to secretion (+0-05+0-05 ml/min/loop) within 20 minutes (p 0-001 or less). Absorption returned to normal after rinsing filtrates out, but the L.T.-induced secretion (+0-11+0-07 ml/min/loop) began 4 hours later. Case-rates The idiopathic-diarrhoea case-rate of trial participants was 84 cases per 10 000 persons/year. Based on the random sample, case-rates for severe (hospitalised) diarrhoea were 7 S.T.+/L.T.+ cases, 4 s.T.+/0 cases and 9 S.T.-/L.T.- cases per 10 000 persons/year. No outpatient S.T.+/0 survey was made but the total S.T.+/L.T. + isolation-rate was 16/10 000/year, which is about half the cholera case attack-rate in the area.1o
Non-vibrao Cholera The clinical findings associated with fxcal S.T.+ /L.T.+ or S.T.+/0 were indistinguishable except in severely ill adult patients with non-vibrio cholera. 70% of these patients had fxcal S.T.+/L.T.+ strains. However, 2 of 20 cases of non-vibrio cholera had S.T.+/0 strains only and 4 cases occurred in patients with no toxigenic
465 TRIAL CASES AND 5 OTHERS
(2 S.T. + /0, 3
S.T. -
/L.T.—)
Patients in shock had absent or barely perceptible radial pulse. Stool volume column gives % in each group whose total stool volume = 0-1 %, 2-4%, or 5% or more of admission body-weight. Pre-admission duration given for acute onset cases only. Post-admission duration given only for number under 12 years old in group. Asterisk indicates 71% of inpatients and 48% of outpatients whose diarrhoea continued after admission. C subs ample of c.H.o. - inpatients. =
-
1118 TABLE 11-CLINICAL DATA AND C.H.O. STATUS OF PATIENTS WITH NON-VIBRIO CHOLERA (NO ANTIBIOTICS): MEAN FIGURES (RANGES IN PARENTHESES)
E. coli, indicating that other unknown xtiological agents exist. Clinical Comparison of Inpatients and Outpatients Outpatients had a history of watery diarrhoea but had mild, very slight or no detectable dehydration. Criteria for admission to hospital included sunken eyes, dry tongue, and decreased skin turgor. A third of adult inpatients were in shock (table I). Admission blood-pressure of children was not measured, but 46% of C.H.O.+ and 74% of c.H.o.- cases had tachycardia over 120 beats/minute. 8% of C.H.O.+ and 14% of C.H.O.- children had weak or undetectable radial pulse. All patients reported multiple watery stools and 80%
reported pre-admission vomiting (except C.H.O.+ outpatients : 50%); vomiting after rehydration was rare. Reported pre-admission duration was between 1 and 38 hours in over 80% of all cases (table I), but among both inpatients (5% of C.H.O.+, 11%of c.H.o.-) and outpatients (14% C.H.O.+, 24% c.H.o.-) durations of 3 to 7 days (79 cases) to 5 weeks (9 cases) were also encountered. Among inpatients only 3% reported bloody stool, usually of long duration, suggesting a non-E. coli xtiology. Abdominal pain was reported by 61 % of C.H.O. + and 40% of C.H.O.- cases, but only 10% reported stool mucus and 5% fever. Fever over 100°F, was found in only 17% of C.H.O.- and 5% of C.H.O.+ inpatients but was absent in non-vibrio cholera cases. Outpatient histories (c.H.o.+%/c.H.o.-%) were not usually positive for stool mucus (19/27), bloody stool (2/6),’ fever (15/7), or abdominal pain (5/17). Fever over 100°F was found in 25% of C.H.O.+ and 11% of c.H.o.- outpatients’ records. Although diarrhoea measurements were inflated owing to frequent mixture with urine, most volumes were very low. Most outpatients had little or no stool during a mean of 6 hours’ observation (range 3-24 hours; table i). In both c.H.o.- and C.H.O.+ inpatients treated with intravenous or oral fluids plus tetracycline, 33% of adults in shock, 58% of adults not in shock, and 50% of all children had no diarrhoea after admission (table I). Mean total stool volume of the rest was only 1.01(all c.H.o.- cases), 1.71(C.H.O.+ adults) and 1.21 (c.H.o.+ children; range 025-48 1). Intravenous fluids were used chiefly for rehydration, and total inpatient needs were low (mean 2-01) because of low stool volumes and use of oral glucosesaline maintenance solutions. Of 75 outpatients who received intravenous rehydration, 72% received the equivalent of less than 5% of body-weight. Mean adult admission weights in all groups were 32-34-5 kg and discharge weights were 1 to 2-2 kg higher. Mean weights of inpatients under 12 years of age were 16-3 kg (C.H.O.+) and 12 kg (c.H.o.-). Mean weight of outpatient children was 10-5 kg (both C.H.O. groups). There were only 2 deaths: one adult (C.H.O.+) was dead on arrival and one child (C.H.O.+) died during treatment.
Discussion Isolation of toxigenic E. coli from diarrhoea does not by itself prove aetiology, but the high percentage of S.T.+/L.T.+ strains in patients with non-vibrio cholera indicates a probable aetiological relationship. It is noteworthy that the S.T.+/L.T.+ isolation-rate is highest in the most severely dehydrated inpatients and is greater among inpatients than among the milder outpatient cases. Furthermore, the dog-intestinal response to L.T. resembles human non-vibrio cholera with respect to diarrhoea rate, duration, and stool composition) There are 27 or more inapparent or mild vibrio cholera cases for every clinically severe case.10 The same may be true of diarrhoea associated with toxigenic E. coli, and the findings of such strains in healthy individuals is not evidence against an xtiological role in diarrhoea patients.11 The S.T.+/0 supernates had the characteristic rapid onset and rinsable s.T. effect in dogs, but had no L.T. effect (4-hour lag, non-rinsable)1 and were negative in the c.H.o. cell assay for L.T. This rules out the presence of L.T. in the S.T.+/0 supernates and is consistent with a recent report of similar strains from Mexico.2 The effect of human E. coli S.T. in dogs has been described previously,5 but the strains used (e.g., CAL 334) apparently produce L.T. also.1 The presence of S.T. + /0 toxigenic E. coli in 33% of a random sample of c.H.o.- inpatients’ isolates suggests that such strains could be one of the commonest causes of idiopathic human diarrhoea in the tropics. L.T. stimulates antitoxin titre rises in some patients, and cross-neutralisation by antisera of different enterotoxins has also been shown;1 12-15 but S.T. apparently induces no antitoxin rise.1 2 The apparent lack of antibody response to S.T.1 2 makes the similarity between s.’r.+/0-associated diarrhoea in man and laboratory animals the only evidence of a possible setiological role at this time.1 s Epidemiological studies are needed to clarify this possibility. Fortunately treatment is similar in enterotoxigenic diarrhoea of different aaiologies. Water and electrolyte replacement, together with appropriate antibiotics, can be given before the ætiological agent is known. In cholera, the clinical response to antibiotics parallels the disappearance of vibrios from the stool and is evidence for the vibrios’ aetiological role. Only 5% of c.H.o.+ E. coli at Dacca are resistant to tetracycline, and a study of the effect of tetracycline on theclinical course of non-vibrio cholera associated with C.H.O.+ E. coli is needed to confirm the causal role of E. coli in this disorder. This study was funded by N. I. H. research agreement no. R27 AI 110048 with the lohns Hopkins Center for Medical Research. The Cholera Research Laboratory is an autonomous organisation created by agreement between the Governments of the People’s Republic of Bangladesh and the United States of America. It receives its basic support from these Governments
1119 and from the Governments of the United Kingdom and Australia. The research described in this publication was done under the above support. We thank the staffs of the Matlab and Dacca Cholera Hospitals and chemistry, bacteriology, and statistics divisions, C.R.L., for their invaluable assistance in this project.
Requests for reprints should be addressed
to
D.R.N.
REFERENCES 1.
Nalin, D. R., Bhattacharjee, A. K., Richardson, S. R. J. infect. Dis
1974, 130, 595. A., Merson, M. H., Wells, J. G., Sack, R. B., Morris, G. K. Lancet, 1975, ii, 239. 3. Smith, H. W., Gyles, C. L.J. med. Microbiol. 1970, 3, 403. 4. Evans, D. J., Jr., Evans, D. G. Silver, R., Gorbach, S. L. Abstracts of the annual meeting of the American Society for Microbiology, 1974, 74, 101. 5. Guerrant, R. L., Brunton, L. L., Schnaitman, T. C., Rebhun, L. I., Gilman, A. G. Infect. Immun. 1974, 10, 320. 6. Dean, A. G., Ching, Y., Williams, R. G., Harden, L. B.J. infect. Dis. 1972, 125, 407. 7. Pierce, N. G., Wallace, C. K. Gastroenterology, 1972, 63, 349. 8. Evans, D. J., Jr., Evans, D. G., Gorbach, S. L. Infect. Immun. 1973, 8, 725. 9. Lindenbaum, J., Benenson, A. S., Greenough III, W. B., Oseasohn, R., Rizvi, S., Saad, A. Lancet, 1965, i, 1081. 10. McCormack, W. M., Islam, S., Fahimuddin, Mosley, W. H. Am. J. Epidem. 1969, 89, 658. 11. Portnoy, B. L., Dupont, H. L. New Engl.J. Med. 1975, 293, 101. 12. Sack, R. B., Jacobs, B., Mitra, R.J. infect. Dis. 1974, 129, 330. 13. Gyles, C. R. ibid. 277. 14. Donta, S. T., Sack, D. A., Wallace, R. B., Dupont, H. L., Sack, R. B. New Engl.J. Med. 1974, 291, 117. 15. Nalin, D. R., Al-Mahmud, A., Curlin, G., Ahmed, A., Peterson, J. Infect. Immun. 1974. 10, 747. 2. Sack, D.
BREAST CANCER IN PATIENTS WITH HASHIMOTO’S THYROIDITIS NOBUHIRO MARUCHI
KUNIHIKO ITOH
Hospital, and Tokyo University School of Medicine, Tokyo, Japan
Itoh Clinic and
Summary
records of four groups of thirty on the books of a a possible associaclinic were used to explore thyroid tion between breast cancer and Hashimoto’s thyroiditis. The index group consisted of 1810 women (10 160 person-years of observation) with Hashimoto’s thyroiditis, and the three control groups, matched by sex, age, marital status, and residency, were selected from patients with myxœdema (essential hypoThe
case
women over
thyroidism), benign
nodular
goitre,
ism. The incidence of breast
or
cancer
hyperthyroidfor the index
though not for control groups, was significantly higher than that expected from data for the general population, suggesting that patients with Hashimoto’s thyroiditis are one high-risk population for breast group,
cancer.
TABLE I-NUMBER OF EXAMINED, PERSON-YEARS OF
N.s.-not
significant.
OBSERVATION,
Introduction THE role of the thyroid gland in breast cancer has been debated for over half a century. 12 That the presence of breast cancer may be associated with thyroid dysfunction or hypothyroidism has been suggested by statistical and pathological evidence,3-6 but several reports do not support these findings,7-11 and thyroid hypofunction as a cause of breast cancer remains controversial. However, with one exception,12 little account has been taken of the incidence of breast cancer in patients with hypothyroidism. The incidence of breast cancer in Japan is about one-fifth of that in the U.S.A. and is almost the lowest among the developed countries,13 so we were rather pessimistic about the practicability of a project that would require the follow-up of more than 1000 index cases and controls. Fortunately, our thyroid registry and case-retrieval system permit the construction of a series for artificial follow-up. The study has revealed more cases of breast cancer in the patients with Hashimoto’s thyroiditis than in control patients such as those with myxoedema, benign nodular goitre, or
hyperthyroidism. Patients and Methods From our thyroid registry we selected as index cases female patients with Hashimoto’s thyroiditis aged over thirty at the time of diagnosis who had been treated for several years for clinical signs and symptoms of Hashimoto’s thyroiditis. A questionnaire was posted to these patients asking about menstrual status, marriage, childbirth, breast feeding, and cancer. Any history of breast tumours was examined in detail with reference to year, name of hospital, diagnosis, and type of treatment. Abstracts were prepared from the medical records and questionnaire, and the time and period of treatment for Hashimoto’s thyroiditis were carefully checked by reviewing the records Three control groups with thyroid diseases were selected: essential hypothyroidism or myxoedema with negative thyroid autoantibodies, benign nodular goitre, and hyperthyroidism. Patients with thyroid cancer were not selected as controls because of another research project on thyroid cancer being undertaken at our institution. Controls were carefully matched for sex, age, marital status, and place of residence with the index group as a whole and information was gathered in the same way as for the index cases. When a patient gave a history of breast tumour, pathological and treatment details were sought and the diagnosis was confirmed. The age distribution of population at risk in each group was computed in terms of the person-years of observation. The expected numbers of breast cancers were calculated from the Osaka tumour registry data for 1970,14 because this registry covers a similar area to ours and because its data are approved by the International Union against Cancer.
AND OBSERVED AND EXPECTED NUMBERS OF BREAST
CANCER, BY STUDY GROUP
Prompted by reports that reovirus-like particles are commonly associated with acute childhood gastroenteritis,6-’o ’6 we searched for evidence of infection by this agent in children not infected with either classic bacterial pathogens or L.T.-positive E. coli. An antigenic relation has been demonstrated between the reoviruslike agent found in stools of children with diarrhoea and a morphologically similar virus, N.C.D.V.24-:U. Immunofluorescent-staining, using primary B.E.K. cell monolayers infected with N.C.D.V., provides a sensitive method for detecting serological responses to infection with the human reovirus-like agent. 18-19 Paired sera from 31 children were tested for immunofluorescent-staining antibodies to N.C.D.v. 35% of children tested showed evidence for a recent infection, a percentage similar to that reported in other studies of infection by the reovirus-like 8 10 19 29 agent in children in hospital with enteritis.7 Our seroconversion data do not prove that the cause of diarrhoea in these children was the reovirus-like agent, since we were unable to perform electron-microscopic studies that would have detected the specific agent in their stools. However, the increase in antibody that coincided with their admission to hospital with enteritis provides strong evidence that most, if not all, of the 11 children had enteritis due to infection with the reoviruslike agent. Indeed, because of the reported lack of complete concordance between serological assays with the human and N.C.D.v. antigens,’9 our results may have underestimated the true frequency of infections caused by the reovirus-like agent in the study group. 9 of the 11 children were admitted to hospital during winter months, a time of year when infections by the reoviruslike agent have been reported to be most common.8 10 29 The results of our one-year surveillance of children admitted to hospital in Boston because of diarrhoea further emphasize the aetiological involvement of the reovirus-like agent. They also suggest that L.T.-positive E. coli may not account for a significant percentage of the previously undiagnosed, endemic childhood cases of diarrhoea in such a setting. Finally, the failure to identify a causal agent in two-thirds of these children indicates the need for continued study of the infectious causes of childhood diarrhoea.
ENTEROTOXIGENIC ESCHERICHIA COLI AND IDIOPATHIC DIARRHŒA IN BANGLADESH D. R. NALIN M. RAHAMAN
Requests for reprints should be addressed to D. S., Children’s Hospital Medical Centre, 300 Longwood Avenue, Boston, Massachusetts 02115, U.S.A. REFERENCES
Melnick, J. L., Blattner, R. J., Stephenson, W. B., Robinson, N. M., Burkhardt, M. A. Am. J. Epidem. 1970, 92, 33. Cramblett, H. G., Siewers, C. M. F. Pediatrics, Springfield, 1965, 35, 885 Gorbach, S. L., Khurana, C. M. New Engl.J. Med. 1972, 287, 791. Sack, R. B., Hirschhorn, N., Brownlee, I., Cash, R. A., Woodward, W. E., Sack, D. A. ibid. 1975, 292, 1041. Rudoy, R. C., Nelson, J. D. Am. J. Dis. Child. 1975, 129, 668. Bishop, R. F., Davidson, G. P., Holmes, I. H., Ruck, B. J. Lancet, 1974, i,
1. Yow, M. D., 2. 3. 4.
5. 6.
149. 7. Flewett, T. H., Davies, H., 1974, 27, 608.
Bryden,
A.
S., Robertson, M. J. J clin. Path.
M. YUNUS G. CURLIN
Cholera Research Laboratory, Johns Hopkins University, International Center for Medical Research, G.P.O. Box no.
128, Dacca, Bangladesh
Fæcal Escherichia coli from Bangalee patients with idiopathic diarrhœa were tested in the Chinese hamster ovary cell (C.H.O.) assay to detect production of thermolabile enterotoxin (L.T.). C.H.O.-positive E. coli produce both L.T. and a thermostable toxin (S.T.). C.H.O.-positive strains were found in 19.2% of all cases and in 70% of the most severely ill patients with non-vibrio cholera. E. coli which produce S.T. alone were identified in a third of the C.H.O.-negative cases. Enterotoxigenic E. coli were found in 55% of inpatients with idiopathic diarrhœa and, if an ætiological role is confirmed, may be one of the commonest causes of tropical diarrhœa.
Summary
Introduction
HUMANS and animals with idiopathic diarrhoea may harbour faecal Escherichia coli which produce both thermostable
8. 9.
10. 11. 12.
13.
14.
We thank the nursing staff and the house staff on division 27 at the C.H.M.C. for their help and cooperation; private physicians who collected convalescent-phase specimens for us; and Dr M. Oxman, Dr T. Thornhill, and Dr A. Smith for their helpful suggestions. L.T.-producing strains of E. coli were supplied by Dr R. B. Hornick (B7A, 334A, and B2C) and Dr R. B. Sack (Nagle 6). Adrenal cells used in the enterotoxin assay were supplied by Dr S. T. Donta and Dr R. B. Sack. This study was supported by a National Institute of Allergy and Infectious Diseases training grant and by the U.S. Army Medical Research and Development Command.
J. C. MCLAUGHLIN
15. 16.
(s.T.) and thermolabile (L.T.) enterotoxins
Middleton, P J., Syzmanski, M. T., Abbott, G. D., Bortolussi, R., Hamilton, J. R. Lancet, 1974, i, 1241. Kapikian, A. Z., Kim, H. W., Wyatt, R. G., Rodriguez, W. J., Ross, S, Cline, W. L., Parrott, R. H., Chanock, R. M. Science, 1974, 185, 1049 Davidson, G. P., Bishop, R. F., Townley, R. R. W., Holmes, I H., Ruck, B. J. Lancet, 1975, i, 242. Sack, R. B., Gorbach, S. L., Banwell, J. G., Jacobs, B , Chatterjee, B. D., Mitra, R. C.J. infect. Dis. 1971,123, 378. Gorbach, S. L., Banwell, J. G., Chatterjee, B. D., Jacobs, B, Sack, R B J. clin. Invest 1971, 50, 881. Dupont, H. L., Formal, S. B., Hornick, R. B., Snvder, M J, Libonati, I P., Sheahan, D. B., LaBrec, E. H., Kalas, J. P New Engl.J Med 1971, 285, 1. Dean, A. G., Ching, Y. C., Williams, R. G., Harden, I.. B. J. infect Dis 1972, 125, 407. Punyashthiti, K., Finkelstein, R. Infect. Immun. 1971, 4, 473. Bishop, R. F., Davidson, G. P., Holmes, I. H., Ruck, B J. Lancet, 1973, ii, 1281.
Flewett, T. H., Bryden, A. S., Davies, H., Woode, G. N., Bridger, J. C. Derrick, J. M. ibid. 1974, ii, 61. 18. Flewett, T. H., Bryden, A. S., Davies, H., Morris, C. A. ibid. 1975, i, 4 19. Kapikian, A. Z., Cline, W. L., Mebus, C. A., Wyatt, R. G, Kalica, A R, James, H. D., VanKirk, D., Chanock, R. M., Kim, H. W. ibid p 1056 20. Edwards, P. R., Ewing, W. H. Identification of Enterobacteriaceæ. Minnea17
polis, 1972. 21. Donta, S. T., Moon, H. W., Whipp, S. C. Science, 1974,183, 334 22. Sack, D. A., Sack, R. B. Infect. Immun. 1975,11, 334 23. Donta, S. T., Sack, D. A., Wallace, R. B., DuPont, H I.., Sack, R B New Engl. J. Med. 1974, 291, 117. 24. Fernelius, A. L., Ritchie, A. E., Classick, L. G., Norman, J O, Mebus, C A. Arch. Virusforsch. 1972, 37, 114. 25. Mebus, C. A., Kono, M., Underdahl, N. R., Twiehaus, M J Can. vet J 1971, 12, 69. 26. Welch, A. B., Twiehaus, M. J. Can. J. comp. Med 1973, 37, 287 27. Dolin, R., Blacklow, N. R., Wyatt, R. G., Sereno, M. M Infect Immun
1972, 6, 958. 28. Mebus, C. A., Underdahl, N. R., Rhodes, M. B., Twiehaus, M. I Res Bull Neb. agric. Exp. Stn. 1969, 233. 29. Bryden, A. S., Davies, H. A., Hadley, R. E, Flewett, T. H, Morris, C A. Oliver, P Lancet, 1975, ii, 241. 30. Chrystie, I. L., Totterdell, B., Baker, M. J., Scopes, J. W., Banatvala, J F ibid. p. 79. 31. Smith, H. W., Gyles, C. L.J. med Microbiol. 1970, 3, 387 32. Evans, D. G , Evans, D. J., Pierce, N. F. Infect. Immun 1973, 7, 873 33. Donta, S. T. Unpublished 34. Sack, R. B., Jacobs, B., Mitra, R.J. infect. Dis. 1974, 129, 330
1117
(S.T.+/L.T.+).1-3 Some animal strains produce S.T. alone (S.T.+/0)3 but human S.T.+/O strains have been reported from only 6 patients.24 S.T.+/L.T.+ strains be identified in the Chinese hamster ovary (c.H.o.) assay5 due to morphological changes induced in c.H.o. cells by L.T. There is no simple s.T. test. The mouseloop test has been used,6 but S.T. may decrease intestinalloop absorption without causing actual fluid accumulation,1-7 so false negatives are best avoided by measuring post-toxin intestinal net water fluxes. In this study the prevalence of S.T.+/O and S.T.+/L.T.+ strains in rectal swab cultures of Bangladesh patients with idiopathic diarrhoea was determined by measuring c.H.o. cell and dog-intestinal water flux responses to E. coli culture can
supernatants. Patients and Methods Between Nov. 10, 1974, and July 8, 1975, all verified cases of idiopathic diarhoea among 92 838 participants in a cholera toxoid vaccine trial were reviewed. Birth dates of the majority were known. All females over 1 year old and all males 1-14 years old were included. Most adult males and infants under 1 were excluded due to trial design. A single diarrhoea hospital serves the censused village trial area (population 167 610). Fæcal E. coli from 96% of trial participant patients whose cultures were negative for Vibrio, Shigella, and Salmonella species (123 inpatients and 370 outpatients) and 52 other idiopathic diarrhoea patients’ E. coli strains were tested for S.T.+/L.T.+ in the c.H.o. assay.s These were classified as C.H.O.+ (= S.T.+/L.T.+) or c.H.o.-. A random group of 20% of c.H.o.- inpatients’ E. coli culture supernates and 24 others (including C.H.O.+ controls) were tested for S.T.+/0 in dogs. Patients’ pre-treatment rectal swabs cultured on MacConkey’s, S. S., and Monsur’s agar showed E. coli but no known pathogens. Patients received intravenous and oral replacement of diarrhœal fluid and electrolyte losses. After cultures were plated, almost all patients received tetracycline by mouth 6-hourly for 5 days (250 mg for adults and 125 mg for children). Ten E. coli colonies from each of 545 patients’ MacConkey’s plates were pooled and stored on blood-agar base slants at room temperature in the dark. E. coli pools were confirmed biochemically. For c.H.o. assay trypticase soy broth with 0-6% yeast extract and 0-2% glucose at pH 8 was inoculated with E. coli and incubated 48 hours at 35 °C without shaking. Supernates were tested after cold centrifu-
gation. supernates (and c.H.o.+ were Evans’ controls) prepared by method8 at pH 8. Heated and unheated 60-100 x concentrated hypotonic (4-50 mosmol/1) supernates were tested for s.T. in adajcent 17 cm dog jejunal loops.1 Heated supernates were placed in a boiling-water bath for 20 minutes and cooled before testing. Each loop received 4.0 ml of concentrate. For
dog-loop
tests
C.H.O.
TABLE I-MEAN CLINICAL DATA ON
Twenty of the pooled isolates tested for S.T.+/L.T.+ and S.T.+/0 were from severely dehydrated adult inpatients at Dacca (chiefly males) with typical non-vibrio cholera.9 These patients had a preadmission duration of less than 1 day, were typically admitted in shock, and had total diarrhoea volumes equal to at least 5% of body-weight. They received intravenous therapy but no antibiotics. Results C.H.O. Assay and Dog-loop Tests Most patients presented with mild watery diarrhoea and vomiting of short duration, often with mild abdominal pain but usually without mucoid or bloody stool. 33% of inpatients’ and 15% of outpatients’ E. coli isolates (19-2% of all cases) were C.H.O.+ Seven of twenty-three randomly selected c.H.o.inpatients’ isolates produced only S.T. S.T. was detected by exposure of 28 dog-jejunal loops to six heated and twenty-two unheated aliquots of supernates from nine strains. 20 minutes later, loop absorption (0.11±0.05 ml/min/loop) changed to secretion (+0-05+0-05 ml/
min/loop;
mean
±S.D., P≤0.001). Absorption
re-
mained normal up to 6 hours after rinsing filtrates out. This response is typical of S.T’! Five control loops with no supernate and twenty-two with non-toxic supernates absorbed -0-08 to -0-10 ml/min/loop for up to 6 hours. In seventeen positive control loops S.T.+/L.T.+ supernates induced a change from absorption (-0.13 +0-05) to secretion (+0-05+0-05 ml/min/loop) within 20 minutes (p 0-001 or less). Absorption returned to normal after rinsing filtrates out, but the L.T.-induced secretion (+0-11+0-07 ml/min/loop) began 4 hours later. Case-rates The idiopathic-diarrhoea case-rate of trial participants was 84 cases per 10 000 persons/year. Based on the random sample, case-rates for severe (hospitalised) diarrhoea were 7 S.T.+/L.T.+ cases, 4 s.T.+/0 cases and 9 S.T.-/L.T.- cases per 10 000 persons/year. No outpatient S.T.+/0 survey was made but the total S.T.+/L.T. + isolation-rate was 16/10 000/year, which is about half the cholera case attack-rate in the area.1o
Non-vibrao Cholera The clinical findings associated with fxcal S.T.+ /L.T.+ or S.T.+/0 were indistinguishable except in severely ill adult patients with non-vibrio cholera. 70% of these patients had fxcal S.T.+/L.T.+ strains. However, 2 of 20 cases of non-vibrio cholera had S.T.+/0 strains only and 4 cases occurred in patients with no toxigenic
465 TRIAL CASES AND 5 OTHERS
(2 S.T. + /0, 3
S.T. -
/L.T.—)
Patients in shock had absent or barely perceptible radial pulse. Stool volume column gives % in each group whose total stool volume = 0-1 %, 2-4%, or 5% or more of admission body-weight. Pre-admission duration given for acute onset cases only. Post-admission duration given only for number under 12 years old in group. Asterisk indicates 71% of inpatients and 48% of outpatients whose diarrhoea continued after admission. C subs ample of c.H.o. - inpatients. =
-
1118 TABLE 11-CLINICAL DATA AND C.H.O. STATUS OF PATIENTS WITH NON-VIBRIO CHOLERA (NO ANTIBIOTICS): MEAN FIGURES (RANGES IN PARENTHESES)
E. coli, indicating that other unknown xtiological agents exist. Clinical Comparison of Inpatients and Outpatients Outpatients had a history of watery diarrhoea but had mild, very slight or no detectable dehydration. Criteria for admission to hospital included sunken eyes, dry tongue, and decreased skin turgor. A third of adult inpatients were in shock (table I). Admission blood-pressure of children was not measured, but 46% of C.H.O.+ and 74% of c.H.o.- cases had tachycardia over 120 beats/minute. 8% of C.H.O.+ and 14% of C.H.O.- children had weak or undetectable radial pulse. All patients reported multiple watery stools and 80%
reported pre-admission vomiting (except C.H.O.+ outpatients : 50%); vomiting after rehydration was rare. Reported pre-admission duration was between 1 and 38 hours in over 80% of all cases (table I), but among both inpatients (5% of C.H.O.+, 11%of c.H.o.-) and outpatients (14% C.H.O.+, 24% c.H.o.-) durations of 3 to 7 days (79 cases) to 5 weeks (9 cases) were also encountered. Among inpatients only 3% reported bloody stool, usually of long duration, suggesting a non-E. coli xtiology. Abdominal pain was reported by 61 % of C.H.O. + and 40% of C.H.O.- cases, but only 10% reported stool mucus and 5% fever. Fever over 100°F, was found in only 17% of C.H.O.- and 5% of C.H.O.+ inpatients but was absent in non-vibrio cholera cases. Outpatient histories (c.H.o.+%/c.H.o.-%) were not usually positive for stool mucus (19/27), bloody stool (2/6),’ fever (15/7), or abdominal pain (5/17). Fever over 100°F was found in 25% of C.H.O.+ and 11% of c.H.o.- outpatients’ records. Although diarrhoea measurements were inflated owing to frequent mixture with urine, most volumes were very low. Most outpatients had little or no stool during a mean of 6 hours’ observation (range 3-24 hours; table i). In both c.H.o.- and C.H.O.+ inpatients treated with intravenous or oral fluids plus tetracycline, 33% of adults in shock, 58% of adults not in shock, and 50% of all children had no diarrhoea after admission (table I). Mean total stool volume of the rest was only 1.01(all c.H.o.- cases), 1.71(C.H.O.+ adults) and 1.21 (c.H.o.+ children; range 025-48 1). Intravenous fluids were used chiefly for rehydration, and total inpatient needs were low (mean 2-01) because of low stool volumes and use of oral glucosesaline maintenance solutions. Of 75 outpatients who received intravenous rehydration, 72% received the equivalent of less than 5% of body-weight. Mean adult admission weights in all groups were 32-34-5 kg and discharge weights were 1 to 2-2 kg higher. Mean weights of inpatients under 12 years of age were 16-3 kg (C.H.O.+) and 12 kg (c.H.o.-). Mean weight of outpatient children was 10-5 kg (both C.H.O. groups). There were only 2 deaths: one adult (C.H.O.+) was dead on arrival and one child (C.H.O.+) died during treatment.
Discussion Isolation of toxigenic E. coli from diarrhoea does not by itself prove aetiology, but the high percentage of S.T.+/L.T.+ strains in patients with non-vibrio cholera indicates a probable aetiological relationship. It is noteworthy that the S.T.+/L.T.+ isolation-rate is highest in the most severely dehydrated inpatients and is greater among inpatients than among the milder outpatient cases. Furthermore, the dog-intestinal response to L.T. resembles human non-vibrio cholera with respect to diarrhoea rate, duration, and stool composition) There are 27 or more inapparent or mild vibrio cholera cases for every clinically severe case.10 The same may be true of diarrhoea associated with toxigenic E. coli, and the findings of such strains in healthy individuals is not evidence against an xtiological role in diarrhoea patients.11 The S.T.+/0 supernates had the characteristic rapid onset and rinsable s.T. effect in dogs, but had no L.T. effect (4-hour lag, non-rinsable)1 and were negative in the c.H.o. cell assay for L.T. This rules out the presence of L.T. in the S.T.+/0 supernates and is consistent with a recent report of similar strains from Mexico.2 The effect of human E. coli S.T. in dogs has been described previously,5 but the strains used (e.g., CAL 334) apparently produce L.T. also.1 The presence of S.T. + /0 toxigenic E. coli in 33% of a random sample of c.H.o.- inpatients’ isolates suggests that such strains could be one of the commonest causes of idiopathic human diarrhoea in the tropics. L.T. stimulates antitoxin titre rises in some patients, and cross-neutralisation by antisera of different enterotoxins has also been shown;1 12-15 but S.T. apparently induces no antitoxin rise.1 2 The apparent lack of antibody response to S.T.1 2 makes the similarity between s.’r.+/0-associated diarrhoea in man and laboratory animals the only evidence of a possible setiological role at this time.1 s Epidemiological studies are needed to clarify this possibility. Fortunately treatment is similar in enterotoxigenic diarrhoea of different aaiologies. Water and electrolyte replacement, together with appropriate antibiotics, can be given before the ætiological agent is known. In cholera, the clinical response to antibiotics parallels the disappearance of vibrios from the stool and is evidence for the vibrios’ aetiological role. Only 5% of c.H.o.+ E. coli at Dacca are resistant to tetracycline, and a study of the effect of tetracycline on theclinical course of non-vibrio cholera associated with C.H.O.+ E. coli is needed to confirm the causal role of E. coli in this disorder. This study was funded by N. I. H. research agreement no. R27 AI 110048 with the lohns Hopkins Center for Medical Research. The Cholera Research Laboratory is an autonomous organisation created by agreement between the Governments of the People’s Republic of Bangladesh and the United States of America. It receives its basic support from these Governments
1119 and from the Governments of the United Kingdom and Australia. The research described in this publication was done under the above support. We thank the staffs of the Matlab and Dacca Cholera Hospitals and chemistry, bacteriology, and statistics divisions, C.R.L., for their invaluable assistance in this project.
Requests for reprints should be addressed
to
D.R.N.
REFERENCES 1.
Nalin, D. R., Bhattacharjee, A. K., Richardson, S. R. J. infect. Dis
1974, 130, 595. A., Merson, M. H., Wells, J. G., Sack, R. B., Morris, G. K. Lancet, 1975, ii, 239. 3. Smith, H. W., Gyles, C. L.J. med. Microbiol. 1970, 3, 403. 4. Evans, D. J., Jr., Evans, D. G. Silver, R., Gorbach, S. L. Abstracts of the annual meeting of the American Society for Microbiology, 1974, 74, 101. 5. Guerrant, R. L., Brunton, L. L., Schnaitman, T. C., Rebhun, L. I., Gilman, A. G. Infect. Immun. 1974, 10, 320. 6. Dean, A. G., Ching, Y., Williams, R. G., Harden, L. B.J. infect. Dis. 1972, 125, 407. 7. Pierce, N. G., Wallace, C. K. Gastroenterology, 1972, 63, 349. 8. Evans, D. J., Jr., Evans, D. G., Gorbach, S. L. Infect. Immun. 1973, 8, 725. 9. Lindenbaum, J., Benenson, A. S., Greenough III, W. B., Oseasohn, R., Rizvi, S., Saad, A. Lancet, 1965, i, 1081. 10. McCormack, W. M., Islam, S., Fahimuddin, Mosley, W. H. Am. J. Epidem. 1969, 89, 658. 11. Portnoy, B. L., Dupont, H. L. New Engl.J. Med. 1975, 293, 101. 12. Sack, R. B., Jacobs, B., Mitra, R.J. infect. Dis. 1974, 129, 330. 13. Gyles, C. R. ibid. 277. 14. Donta, S. T., Sack, D. A., Wallace, R. B., Dupont, H. L., Sack, R. B. New Engl.J. Med. 1974, 291, 117. 15. Nalin, D. R., Al-Mahmud, A., Curlin, G., Ahmed, A., Peterson, J. Infect. Immun. 1974. 10, 747. 2. Sack, D.
BREAST CANCER IN PATIENTS WITH HASHIMOTO’S THYROIDITIS NOBUHIRO MARUCHI
KUNIHIKO ITOH
Hospital, and Tokyo University School of Medicine, Tokyo, Japan
Itoh Clinic and
Summary
records of four groups of thirty on the books of a a possible associaclinic were used to explore thyroid tion between breast cancer and Hashimoto’s thyroiditis. The index group consisted of 1810 women (10 160 person-years of observation) with Hashimoto’s thyroiditis, and the three control groups, matched by sex, age, marital status, and residency, were selected from patients with myxœdema (essential hypoThe
case
women over
thyroidism), benign
nodular
goitre,
ism. The incidence of breast
or
cancer
hyperthyroidfor the index
though not for control groups, was significantly higher than that expected from data for the general population, suggesting that patients with Hashimoto’s thyroiditis are one high-risk population for breast group,
cancer.
TABLE I-NUMBER OF EXAMINED, PERSON-YEARS OF
N.s.-not
significant.
OBSERVATION,
Introduction THE role of the thyroid gland in breast cancer has been debated for over half a century. 12 That the presence of breast cancer may be associated with thyroid dysfunction or hypothyroidism has been suggested by statistical and pathological evidence,3-6 but several reports do not support these findings,7-11 and thyroid hypofunction as a cause of breast cancer remains controversial. However, with one exception,12 little account has been taken of the incidence of breast cancer in patients with hypothyroidism. The incidence of breast cancer in Japan is about one-fifth of that in the U.S.A. and is almost the lowest among the developed countries,13 so we were rather pessimistic about the practicability of a project that would require the follow-up of more than 1000 index cases and controls. Fortunately, our thyroid registry and case-retrieval system permit the construction of a series for artificial follow-up. The study has revealed more cases of breast cancer in the patients with Hashimoto’s thyroiditis than in control patients such as those with myxoedema, benign nodular goitre, or
hyperthyroidism. Patients and Methods From our thyroid registry we selected as index cases female patients with Hashimoto’s thyroiditis aged over thirty at the time of diagnosis who had been treated for several years for clinical signs and symptoms of Hashimoto’s thyroiditis. A questionnaire was posted to these patients asking about menstrual status, marriage, childbirth, breast feeding, and cancer. Any history of breast tumours was examined in detail with reference to year, name of hospital, diagnosis, and type of treatment. Abstracts were prepared from the medical records and questionnaire, and the time and period of treatment for Hashimoto’s thyroiditis were carefully checked by reviewing the records Three control groups with thyroid diseases were selected: essential hypothyroidism or myxoedema with negative thyroid autoantibodies, benign nodular goitre, and hyperthyroidism. Patients with thyroid cancer were not selected as controls because of another research project on thyroid cancer being undertaken at our institution. Controls were carefully matched for sex, age, marital status, and place of residence with the index group as a whole and information was gathered in the same way as for the index cases. When a patient gave a history of breast tumour, pathological and treatment details were sought and the diagnosis was confirmed. The age distribution of population at risk in each group was computed in terms of the person-years of observation. The expected numbers of breast cancers were calculated from the Osaka tumour registry data for 1970,14 because this registry covers a similar area to ours and because its data are approved by the International Union against Cancer.
AND OBSERVED AND EXPECTED NUMBERS OF BREAST
CANCER, BY STUDY GROUP
