Breast cancer survival among women under age 50: is mammography detrimental?

991

5. McDonald WI. The mystery of the origin of multiple sclerosis. J Neurol Neurosurg Psychiatry 1986; 49: 434-45. 6. Compston AC. Genetic factors in the aetiology of multiple sclerosis. In: McDonald WI, Silberberg DH, eds. Multiple sclerosis. London: Butterworth, 1986: 56-73. 7. Ebers GC, Paty DW, Stiller C, Nelson R, Seland T. HLA typing in MS sib pairs. Lancet 1982; ii: 88-90. 8. McFarland HF, Dhib-Jalbut S. Multiple sclerosis: possible immunological mechanisms. Clin Immunol Immunopathol 1989; 50:

S96-S105. 9. Alvord EC Jr, Kies MW, Suckling AJ, eds. Experimental allergic

encephalomyelitis: a useful model for multiple sclerosis. New York: Alan R. Liss, 1984. 10. Boylan KB, Takahashi N, Diamond M, Hood LE, Prusiner S. DNA length polymorphism located 5’ to the human myelin basic protein gene. Am J Hum Genet 1987; 40: 387-400. 11. Boylan KB, Ayers TM, Popko B, Takahashi N, Hood LE, Prusiner S. Repetitive DNA (TGGA)n 5’ to the human myelin basic protein gene: a new form of oligonucleotide repetitive sequence showing length polymorphism. Genomics 1990; 6: 16-22. 12. Boylan KB, Takahashi N, Paty DW, et al. DNA length polymorphism 5’ to the myelin basic protein gene is associated with multiple sclerosis. Ann Neurol 1990; 27: 291-97. 13. Nevanlinna H. The Finnish population structure: a genetic and genealogical study. Hereditas 1972; 71: 195-236. 14. Poser CM, Paty DW, Scheinberg L, et al. New diagnostic criteria for multiple sclerosis: guidelines for research protocols. Ann Neurol 1983; 13: 227-31. 15. Tienari PJ, Salonen O, Wikström J, Valanne L, Palo J. Familial multiple sclerosis: MRI findings in clinically affected and unaffected siblings. J Neurol Neurosurg Psychiatry (in press). 16. Paty DW, Oger JJG, Kastrukoff LF, et al. MRI in the diagnosis of MS. Neurology 1988; 38: 180-85. F, Offenbacher H, Fuchs S, et al. Criteria for an increased specificity of MRI interpretations in elderly subjects with suspected multiple sclerosis. Neurology 1988; 38: 1822-25.

17. Fazekas

Breast

survival among women under age 50: is mammography detrimental?

breast cancer by mammography in under 50 years of age. We have reviewed the survival of patients aged 49 years or less whose cancers were detected by mammography alone. 117 women under the age of 50 years were diagnosed with breast cancer between 1978 and 1991 based only on an abnormal mammogram. Ductal carcinoma in-situ (DCIS) was found in 47 (40%) of these women, whilst 70 (60%) had infiltrating ductal or infiltrating lobular carcinomas. During the same interval, 928 women in this age group presented with palpable breast cancer. DCIS was diagnosed in 82 (9%) of these women, whilst 846 (91%) had infiltrating carcinoma. Among the infiltrating cancers detected by mammography alone, 50% were stage I, whilst only 30% of the women with palpable cancers were stage I. Fiveyear survival for all mammographically detected cancer patients was 95%, whereas for women with palpable cancers the survival was 74% If DCIS is not included, the (p<0·00005).

detecting

1990; 35: 1196-200. 19. Sambrook J, Fritsch EF, Maniatis T. Molecular cloning. New York: Cold Spring Harbor Laboratory Press, 1989. 20. Saiki RK, Gelfand DH, Stoffel S, et al. Primer-directed enzymatic amplification of DNA with a thermostable DNA polymerase. Science 1988; 239: 487-91. 21. Sajantila A, Budowle B, Ström M, et al. PCR amplification of alleles at the D1S80 locus: comparison of a Finnish and north American caucasian population sample, and forensic casework evaluation. Am J Hum Genet 1992; 50: 816-25. 22. Kamholtz J, Spielman R, Gogolin K, Modi W, O’Brien S, Lazzarini R. The human myelin basic protein gene: chromosomal localization and RFLP analysis. Am J Hum Genet 1987; 40: 365-73. 23. Ott J. Analysis of human genetic linkage. Baltimore: Johns Hopkins University Press, 1991. 24. Visscher B, Detels R, Dudley JP, et al. Genetic susceptibility to multiple sclerosis. Neurology 1979; 29: 1354-60. 25. Sandberg-Wollheim M, Bynke H, Cronqvist S, et al. A long term prospective study of optic neuritis: evaluation of risk factors. Ann Neurol 1990; 27: 386-93. 26. Rizzo JF, Lessell S. Risk of developing multiple sclerosis after uncomplicated optic neuritis: a long term prospective study. Neurology 1988; 38: 185-90. 27. Roach A, Takahashi N, Pravtcheva D, Ruddle F, Hood LE. Chromosomal mapping of mouse myelin basic protein gene and structure and transcription of the partially deleted gene in shiverer mutant mice. Cell 1985; 42: 149-55. 28. Okano H, Tamura T, Miura M, et al. Gene organization and transcription of duplicated MBP genes of myelin deficient (shimld) mutant mouse. EMBO J 1988; 7: 77-83. 29. Fujinami RS, Oldstone MBA. Amino acid homology between the encephalitogenic site of myelin basic protein and virus: mechanism for autoimmunity. Science 1985; 230: 1043-45.

cancer

Great uncertainty exists about the benefit of women

K, Nakamura Y, White R. Amplification of a variable number of tandem repeats (VNTR) locus (pMCT118) by the polymerase chain reaction (PCR) and its application to forensic science. J Forensic Sci

18. Kasai

survivals are for 91% corresponding mammographically detected infiltrating cancers and 72% for palpable infiltrating cancers. Only 1 woman who died among those with palpable cancer had had a mammogram before diagnosis. Our data contradict the suggestion that women under 50 are put at a survival disadvantage by undergoing mammography. We believe that investigators who have reported negative results in

this age group must examine other results.

causes

for their

Lancet 1992; 340: 991-94.

ADDRESSES: Department of Radiology, Massachusetts General Hospital; Massachusetts General Hospital Cancer Center; and Harvard Medical School, Boston, Massachusetts (K. A McCarthy, MD, D. A. Hall, MD, E. Pile-Spellman, MD, G. White, MD, C. Hulka, MD, G J Whitman, MD, D. B Kopans, MD); Division of Surgical Oncology, Harvard Medical School, Massachusetts General Hospital (A Stacey-Clear, FRCS); and Department of Biostatistics, Harvard School of Public Health, Boston, Massachusetts (E Mahoney, MS) Correspondence to Dr D. B Kopans, Department of Radiology, Massachusetts General Hospital, Boston, Massachusetts 02114, USA.

992

Introduction There is great controversy about the benefit of detecting breast cancer earlier by mammography in women under the age of 50 years. None of the large randomised controlled trials reported to date have been designed to assess these women separately.1-3 They have all set out to measure benefit when screening is begun at age 40 with no stratification, and most of these trials have lacked sufficient numbers of women in the age group 40-49 to have the statistical power to detect such a benefit. Meta-analysis is unlikely to help because the design and execution of these studies were not planned specifically for the 40-49 year age group. For instance, in the Swedish two-county trial, most women under 50 were screened every 2 years. Among women aged 40-49, mammography can detect breast cancer about 2 years before the tumour becomes clinically evident.’ If a 2 year or longer time is taken between screens, the cancers will be detected at about the time the woman or her physician would feel a mass in the absence of mammographic screening, and little gain will be achieved from the screen. If mammography is used to screen women aged 40-49, they should be screened with an interval (time between screens) of no longer than 1 year. The two-county trial also screened with single-view oblique mammography, and this has been shown to reduce the detection rate of breast cancer.s Interpretation of trial results is further complicated by the fact that women who do not take part in screening are counted as having been screened. Although this manoeuvre eliminates selection bias, cancers developing in these women dilute the magnitude of a screening benefit. A further difficulty has been the crossover of control patients who availed themselves of mammographic screening outside the trial. Despite having been screened, these women are still counted as unscreened controls. In the Malmo study, 35 % of younger women who were assigned as controls underwent mammography outside the trial, and at least 20% of the control cancers were detected by mammography.6 The power to detect a benefit is diminished when controls are screened. The Canadian National Breast Cancer Screening Trial (NBSS) was in part designed to assess the benefit of screening women under 50. Although the results have not yet been fully published, data have been presented at various meetings and these suggest that paradoxically there were more deaths among the screened population between the ages 40 and 49 than among controls. The publication of these incomplete results has caused widespread concern.7 However, this unexpected and counterintuitive result has not been confined to the Canadian study. A similar finding was observed in the early years after the onset of screening in other randomised trials,8 but the magnitude of the difference may be much larger in the NBSS. The organisers of the NBSS have suggested that mammography might in some way be deleterious to survival, and other workers have also suggested this possibility.9 We have reviewed the survival of our patients aged 49 or younger, whose cancers were detected by mammography alone, and compared them with women whose breast cancers were diagnosed without the benefit of mammography.

Patients and methods All women aged under 50 years who were diagnosed with breast cancer at the Massachusetts General Hospital between May, 1978, and April, 1991, were included in our study. The survival of women whose breast cancer was detected by mammography alone and who required needle localisation to guide the surgeon was compared

women under 50 whose cancers were diagnosed because of a clinically apparent abnormality. The date of diagnosis was taken as the date of needle localisation and biopsy for the mammographically detected cancer, and the date of biopsy was assigned as the date of diagnosis for the clinically apparent cancer. Survival was calculated from these dates. Patient data were obtained from their charts, the radiology breast imaging computer records, and hospital tumourregistry records. Follow-up information in the tumour registry was obtained from analysis of repeat visits to the hospital and subsequent correspondence between the tumour registry, physicians, and other area hospitals.

with

ductal carcinoma in-situ (DCIS), infiltrating lobular carcinoma. Lobular carcinoma in-situ was not included because of its uncertain biological importance. Hospital tumour registry material was used to establish survival periods. Tumour registry follow-up has a six-month lag with a 90% rate of follow-up success. Because of inaccuracies in the measurement of the extent of DCIS, the sizes of tumours include only invasive lesions. The size of these cancers was measured as the greatest diameter of the tumour mass (spicules beyond the tumour mass were not included). If lymph-node dissection was done, the number of positive nodes was recorded. Staging was made according to TNM criteria. Stage 2 or worse tumours were those whose diameter was 2 cm or more, or a tumour of any size with axillary nodal involvement or other evidence of metastatic spread. Breast

cancer was

defmed

as

infiltrating ductal carcinoma,

or

Results There

under 50 years of age whose the 156 month period. Among these women, 117 had their cancers detected by mammography, whilst 928 presented with clinical evidence of an abnormality. were

cancers were

1045

women

diagnosed during

Mammographically detected lesions whose cancers were detected by a had mean age of 43-5 years (median 45, mammography At least of these women had had at least 36% range 29-49). one mammogram before detection and diagnosis of cancer. The follow-up interval ranged from 15 to 166 months (mean 65, median 54). Among the 70 women in this group with invasive cancers, the follow-up was 15 to 166 months, with 29 women followed for a minimum of 5 years. 7 patients were lost to follow-up (range 1 week-4 years). Of the 117 mammographically detected lesions, 70 (60%) patients had invasive carcinoma (69 invasive ductal carcinoma and 1 invasive lobular carcinoma), and 47 (40%) women had DCIS. Invasive tumours ranged in size from 0-15 cm to 6-0 cm (mean 1-45, median 1-20). Among the 70 women with invasive cancer, 28 (40%) tumours were 1 em or smaller in size. 21 women had positive lymph nodes. Based on size and/or node positivity, 35 (50%) patients whose invasive cancers were detected by mammography alone were stage 2 or worse. If women with DCIS are included, then the proportion of stage 2 or worse lesions falls to 30%. There have been seven deaths among the women whose cancers were detected by mammography. 1 patient had in-situ breast cancer detected mammographically, but she died thirteen months later from a primary carcinoma of the bronchus. At the time of death she had no evidence of recurrent breast cancer. The other six deaths were due to breast cancer. 2 of these women had a mammographically detected lesion simultaneously with a separate palpable cancer, but they have been included with the mammographically detected lesions. Among the other four cancer deaths in the mammographically detected group, 1 patient had diffuse breast involvement with DCIS and microscopic invasion at the time of diagnosis. Another

The 117

women

993

0

1 2 3 Years since diagnosis

5 4

Fig 1-Five-year actuarial survival for clinically mammographically detected breast cancer. . Mammographically detected. 0 Clinically detected.

and

patient developed a metachronous contralateral cancer one year later with positive lymph nodes. The 2 remaining patients with mammographically detected cancers had positive lymph nodes at the time of diagnosis. Thus, four of seven deaths in this group were among women with advanced disease at the time of breast cancer diagnosis (2 women with synchronous palpable masses and 2 women with positive lymph nodes).

Clinically apparent breast cancer Because of the uncertainties about the benefits of mammography for women under age 50, mammographic screening was not practised commonly for women in this age group before 1990. Most women aged under 49 years with cancer presented with clinical abnormalities. Thus, 928 patients in this group who were diagnosed during the same interval with breast cancer had clinically apparent abnormalities. Among these, 82 (9%) were diagnosed with DCIS, whilst the remaining 846 had infiltrating cancer. Of women presenting with clinically evident breast cancer, 412 had positive axillary lymph nodes at the time of diagnosis. Stage analysis revealed that, among the clinically apparent cancers, 595 were stage 2 or worse at the time of diagnosis (70% of the infiltrating cancers; 64% if DCIS is included). 256 women have died. 148 deaths were attributed to breast cancer for clinically evident tumours. In addition, there have been 64 deaths from an unestablished cause but with residual or recurrent cancer present. 36 women have died with an unestablished breast cancer status and with undocumented causes. 5 patients have died free of breast 1 nn

-

0

1

2

_

3

4

Years since

Fig 2-Ten-year actuarial mammographically detected . Mammographically detected. 0

Clinically detected.

5

6

7

9 8

10

diagnosis

survival for breast cancer.

clinically

and

disease, and 3 patients have died from other causes but with breast cancer present. We have reviewed the breast imaging information on women with clinically evident abnormalities who have died of breast cancer and, of the 180 women whose mammographic records could be retrieved, only 1 had had a mammogram before presentation with a clinical abnormality (no abnormality was evident in this one case). The five-year survival rate (all causes of death) for patients under 50 with breast cancer was 77%. The survival rate for patients under 50 with mammographically detected cancer was 95% (DCIS included). IfDCIS is excluded, the survival is 91 %. This figure was compared with a five-year survival rate of 74% for clinically evident breast cancer (72% if only invasive cancer is counted). Actuarial survival curves at five and ten years for both mammographically detected and clinically evident cases are shown in figs 1 and 2. Women with mammographically detected cancers had a significantly better prognosis for survival than did those women with palpable disease (p < 0-00005, log rank test). These results are not qualitatively altered if data are adjusted to assume a two-year lead time for mammographic detection. Discussion The incidence of breast

cancer

in the USA has been

increasing steadily during the past 50 years,1O with a recent dramatic upturn in the past 5 years." Much, although not all, of the recent additional increase has been artificial because of earlier detection by mammography. Breast cancer has been overtaken by lung cancer as the leading of cancer death among US women, but it remains the leading cause of non-preventable cancer death. Despite this increase in incidence, the death rate from the disease has remained fairly constant. This may suggest an improvement in mortality, but few observers will be convinced until an obvious decrease in mortality is reported. If mammographic screening is to accomplish this task, a large percentage of the population will have to be screened. There is an understandable reluctance to endorse population screening without absolute proof of benefit. The controversy about screening women under 50 is in part economic. In a computer analysis of data from trials that cause

designed specifically to study women aged 40-49, Eddy et al found that there was a small benefit from screening women under 50 but that it was too costly to be justified.12 There are, as yet, no data from randomised controlled trials that can be used to examine accurately the benefit of mammography in these younger women. Health planners are awaiting the results of the NBSS to make important decisions. Our data and those previously published,13 together with the questions that have been raised about the design and execution of the NBSS, cast serious doubt on the validity of the results that will emerge were not

from that trial. Several explanations have been cited by the principal investigators of the NBSS to account for excess cancer deaths among screened women, including a detrimental effect from the mammography itself. Mechanisms ranging from the potential adverse effects of compression9 to the radiographic potentiation of cancers have been suggested. The most likely reasons for the observation of excess mortality in screening programmes have been discounted. Poor-quality mammography is an important component of false-negative mammography. 14 False-negative mammography contributes to delayed diagnosis and a

994

higher stage at diagnosis in symptomatic women. This effect is exaggerated by poor-quality mammography. Excess mortality from this effect can be caused in symptom-free women undergoing screening by false reassurance. A woman who is screened with negative results may be reassured erroneously by the negative mammogram. She might be more likely to delay seeking help if she noticed a change in her breast subsequent to the negative screen. By comparison, her control counterpart would be more likely to seek immediate advice on finding a change in her breast. The control subject would therefore be diagnosed earlier than the screened woman and this would lead to an excess mortality among screened women. Poor-quality mammography would exacerbate this negative effect by falsely reassuring women and delaying the diagnosis and treatment of cancers. This effect would reverse the benefit that might accrue from high-quality mammography. 15 There is evidence that women in screening programmes are aware of breast changes but await their next screen rather than seek immediate evaluation. 16 The survival of patients with breast cancer is directly related to stage at diagnosis. Staging is, of necessity, artificial, and there are subpopulations within stages who have different prognoses. Rosen et al have shown that among stage I cancers stratification by size reveals a survival advantage for the Tl NO MO patients whose tumours were 1 -0 cm or less in diameter. These women had a significantly better 20 year recurrence-free survival (86%) than did Tl1 NO MO patients with tumours 11-20 cm (69%).18 Tabar and colleagues have recently reported a similar benefit from the detection of cancers below 1 em.19 Thus, the large difference in proportion of patients with stage 2 or worse disease (and that of DCIS) between mammographically detected and clinically detected cases cannot be ignored. Survival data cannot be used to conclusively demonstrate a benefit for the mammographic screening of women aged 40-49. Our study was retrospective and without randomisation, the number of purely mammographically detected cancers was small, and a full five-year follow-up only accounted for 43% of the cases of invasive cancer. However, some inferences about the likely benefit of screening can be made from survival data such as ours. Our results cast doubt on data from trials which show no benefit for screening women under 50. Selection bias, lead-time bias, and length-bias sampling are all important difficulties, but the results of randomised controlled trials (eg, the Swedish two-county trial) have shown that a reduction in size and stage at the time of diagnosis is the main reason for true mortality reduction. Survival data therefore become a valid measure if they relate directly to size and stage. Our data confirm that it is more difficult to detect breast cancers earlier in younger women. Although the stage at diagnosis among these women is reduced compared with women who have palpable cancers, the stage at diagnosis for mammographically detected cancer is still higher among younger women than for older women." Nevertheless, our review indicates that mammography can still detect these lesions earlier than physical examination. The ability to detect breast cancer at a smaller size and at an earlier stage at ages both above and below 50, together with the direct effects of these factors on prognosis, raises important questions about screening trials that fail to show a benefit. Detailed review of the design and execution of these trials must be made to fully understand their results. We have deliberately reported our results both including and excluding DCIS because of the lack of agreement as to

the natural history of this lesion. The importance of intraductal carcinoma in the pathogenesis of invasive cancer remains unclear. Given time, a high proportion of these lesions will probably become malignant. Thus, DCIS in the population of women under 50 years is likely to be an important lesion, but the effect of finding it may not be measurable for 10-15 years. An additional question that must be taken into account when evaluating results from randomised controlled rials is the threshold used for intervening and obtaining tissue diagnosis. In the Swedish two-county trial, which has not as yet shown a mortality reduction for women under 50, the positive predictive value for mammographically instigated biopsy was 50-70%.21 In our study, the positive predictive value for women under 50 was 12%. Our results suggest that a more aggressive diagnostic approach is needed in younger women.22 REFERENCES Venet W, Strax P, Venet L. Periodic screening for breast the health insurance plan project and its sequelae, 1963-1986. Baltimore: Johns Hopkins University Press, 1988. 2. Tabar L, Gad A, Holmberg LH, et al. Reduction in mortality from breast cancer after mass screening with mammography. Lancet 1985; i: 829-32. 3. Andersson I, Aspegren K, Janzon 1, et al. Mammographic screening and mortality from breast cancer: the Malmo mammographic screening trial. BMJ 1988; 297: 943-49. 4. Tabar L, Faberberg G, Day NE, Holmberg L. What is the optimum interval between mammographic screening examinations? An analysis based on the latest results of the Swedish two-county breast screening trial. Br J Cancer 1989; 55: 547-51. 5. Muir BB, Kirkpatrick A, Roberts MM, Duffy SW. Oblique-view mammography: adequacy for screening. Radiology 1984; 151: 39-41. 6. Andersson I, Aspegren K, Janzon L, et al. Mammographic screening and mortality from breast cancer: the Malmo mammographic screening trial. BMJ 1988; 297: 943-49. 7. Editorial. Breast cancer screening in women under 50. Lancet 1991; 337: 1575-76. 8. Stomper PC, Gelman RS. Mammography in symptomatic and asymptomatic patients. Hematol Oncol Clin N Am 1989; 3: 611-40. 9. Watmough DJ, Quan KM. X-ray mammography and breast compression. Lancet 1992; 340: 122. 10. White E, Lee CY, Kristal AR. Evaluation of the increase in breast cancer incidence in relation to mammography use. J Natl Cancer Inst 1990; 82: 1546-52. 11. Miller BA, Feuer EJ, Hankey BF. Trends in invasive breast cancer incidence among American women. J Natl Cancer Inst 1991; 83: 678. 12. Eddy DM, Hasselblad V, McGivney W, Hendee W. The value of mammography screening in women under 50 years. JAMA 1988; 259: 1512-19. 13. Seidman H, Gelb SK, Silverberg E, La Verda N, Lubera JA. Survival experience in the breast cancer detection demonstration project. CA 1987; 37: 258-90. 14. Martin J, Moskowitz M, Milbrath JR. Breast cancers missed by mammography. AJR 1979; 132: 737-39. 15. Moskowitz M. Guidelines for screening for breast cancer: is a revision in order? In: Breast imaging: current status and future directions. Radiol Clin N Am 1992; 30: 221-33. 16. Joensu H, Klemi PJ, Tuominen J, Rasanen O, Parvinen I. Breast cancer found at screening and previous detection by women themselves. Lancet 1992; 339: 315. 17. Carter CL, Allen C, Henson DE. Relation of tumor size, lymph node status, and survival in 24,740 breast cancer cases. Cancer 1989; 63: 181-87. 18. Rosen PP, Groshen S, Saigo PE, Kinne DW, Hellman S. A long-term follow-up study of survival in stage I (T1 N0 M0) and stage II (T1 N1 M0) breast carcinoma. J Clin Oncol 1989; 7: 355-66. 19. Tabar L, Fagerberg G, Day N, Duffy SW, Kitchin RM. Breast cancer treatment and natural history: new insights from results of screening. Lancet 1992; 339: 412-14. 20. Kopans DB, Screening mammography in women over age 65 (in press). 21. Tabar L, Gad A, Holmberg L, Ljunquist U. Significant reduction in advanced breast cancer. Diag Imag Clin Med 1985; 54: 158-64. 22. Moskowitz M, Gartside PS. Evidence of breast cancer mortality reduction: aggressive screening in women under age 50. AJR 1982; 138: 911-16. 1.

Shapiro S, cancer: