| Japanese Journal of Clinical Oncology | Pages |
Primary and Adjuvant Therapy, Prognostic Factors and Survival in 1053 Breast Cancers Diagnosed in a Trial of Mammography Screening
Introduction
Materials And Methods
Results
Discussion
References
Primary and Adjuvant Therapy, Prognostic Factors and Survival in 1053 Breast Cancers Diagnosed in a Trial of Mammography Screening
Background: As mammographic screening becomes more widespread, larger numbers of tumours are diagnosed while small and node negative. Methods: We examined detection mode, tumour size, node status, histological type, therapy and outcome in 1053 breast cancers diagnosed in one county of the Swedish Two-County Trial of mammographic screening for breast cancer. Results: Of patients undergoing total mastectomy with axillary dissection, 65% were found to be node negative. For tumours of size 1-9 mm, 95% were node negative. The major effects on survival were tumour size, node status and histological type. Primary adjuvant therapy had no significant association with survival. Conclusions: The advent of mammography has substantially enhanced the possibilities for less radical treatment. There is an urgent need for therapeutic trials utilizing mammographic-pathological correlations to ascertain in advance which tumours can and which cannot benefit from more radical therapy.
Keywords: breast cancer - prognostic factors - breast cancer screening - therapy
INTRODUCTION
During the last two decades, therapy for breast cancer has tended toward less radical surgery and more adjuvant systematic therapy. Clinical trials have indicated that breast-conserving surgery with axillary dissection and radiotherapy is as effective as radical mastectomy for tumours less than 2 cm (1). Indeed, it has been proposed that in tumours with a favourable prognosis, systematic therapy is unnecesssary (2,3).
The main reason for the change in surgical practice from routine mastectomy to breast-conserving surgery or dispensing with routine radiation therapy is the decrease in tumour size and in the proportion of tumours with lymph node involvement, which has been a feature of the era of mammography screening. The observations of Cadyet al. (4) suggest that in this era, 33% of diagnosed cancers will be in situ and of the invasive tumours, the median size is likely to be around 1 cm.
Although clinical trials of adjuvant systematic therapy indicate a benefit in survival terms associated with chemotherapy, it is not clear whether this benefit applies to cancers which already have a very good prognosis. In an overview of chemotherapy for `early' breast cancer, beneficial effects were noted for both hormone and cytotoxic chemotherapy (5). The definition of early, however, included tumours of size up to 5 cm, a definition which is clearly inappropriate to the age of mammographic screening (6).
There is now overwhelming evidence that breast cancer screening with mammography reduces mortality from the disease (7). This implies that breast cancer is a progressive disease, the development of which can be arrested by screening. The substantial difference in survival by tumour attributes such as size, node status and malignancy grade (greater than can be achieved by variation in therapy) indicates that the stage at which progression is halted is crucial (8).
It has been shown that a large part of the benefit of screening is via its reduction of the rate of advanced tumours in terms of size and lymph node status (7,8). Results also indicate that intratumour heterogeneity with respect to measures of aggressive potential is a commonly observed phenomenon (9). In such cases, the more poorly differentiated component (i.e. the more aggressive) grows more rapidly than the better differentiated (10). Thus, early detection can actually improve the overall malignancy grade of a tumour and indeed this was observed to occur in the Swedish Two-County study (8,10).
In summary, the use of mammography screening for breast cancer has established a tumour population in which there are substantial numbers of node-negative tumours, tumours of size less than 1 cm and well-differentiated or moderately differentiated cancers. In addition, certain histological types, such as tubular carcinoma, are known to have an excellent long-term prognosis (10). It has long been axiomatic that early detection confers an opportunity for use of less radical treatment as well as a reduction in mortality and indeed the use of routine adjuvant therapy has been found to be unwarranted in small, node-negative tumours (11).
The Swedish Two-County Trial is a randomized trial of mammographic screening for breast cancer, in which a substantial proportion of tumours less than 1 cm in maximum diameter were diagnosed. In this study, we propose to:
- describe the treatments administered by histology, tumour size and lymph node status;
- ascertain how these are associated with long term outcome; and
- make some tentative suggestions about which tumours may benefit from more radical therapies and which may not.
It should be noted that this paper does not report on a therapeutic trial. The tumours we report on were diagnosed during a randomized trial of mammographic screening. To avoid confounding by treatment in this trial, it was necessary to have the same therapeutic policy throughout, depending only on the stage of the disease and not varying by arm of the screening trial. The results should, however, have implications for future therapeutic trials.
MATERIALS AND METHODS
Treatment data were available from one county (Kopparberg) of the Swedish Two-County Trial, which was a randomized breast cancer screening trial initiated in 1977. A detailed description of this study was published elsewhere (12). In Kopparberg county, a total of 1053 breast cancers were diagnosed, 694 in the invited group (active study population, ASP) and 359 in women not invited (passive study population, PSP). The average follow-up until 31 December 1994 is 14.8 years. In Kopparberg county, deaths from breast cancer were 140 among 38,589 women in the ASP and 113 among 18,582 in the PSP, respectively (RR = 0.60, 95% CI 0.46-0.77).
Basic information on cancers diagnosed include tumour size, node status, malignancy grade and histological type. Detection modes include prevalent (first) screen, incidence (subsequent) screen, interval cancers (cancers diagnosed between screens), cancers in refusers, cancers in the control group diagnosed before screening, cancers in the control group diagnosed at first screen (in this study, the control group was screened once at the end of the trial).
Data on primary and adjuvant therapy were collected for each subject, by filling in a treatment form, constructed by the trialist, at primary treatment and at each routine follow-up examination (follow-up examinations took place four times during the first postoperative year, three times during the second, biennially thereafter up to year five and annually thereafter). Primary therapy consisted of 203 quadrantectomies (QM), five sub-mastectomies (SM), 46 total mastectomies without axillary dissection (TMNA), 693 total mastectomies with axillary dissection (TMA), 17 radical mastectomies (RM), four needle biopsies, 13 no surgery and 72 unknown (see Table 1). Systematic adjuvant therapy was divided into two types. The first adjuvant therapy was prophylactic, administered immediately after surgery. Prophylactic adjuvant therapy included radiotherapy, chemotherapy, hormone therapy and combinations of these. The second was for therapeutic purposes. After primary and prophylactic adjuvant therapy, various adjuvant therapies (referred to as `therapeutic adjuvant therapy' here) were given at different follow-up times according to patient responses (e.g. in response to recurrence) after primary and prophylactic therapy. Interpretation of effects of therapy in response to recurrence is complicated by the fact that both mode of treatment and subsequent response are dependent on the time, site and mode of recurrence. We therefore do not present results with respect to therapeutic adjuvant therapy. We do, however, show the results with respect to other factors adjusted for therapeutic adjuvant therapy (taking into account time to such therapy).
Table 1. Numbers of breast cancers (deaths/cases) cross-tabulated by primary and prophylactic adjuvant therapies for women aged 40-74 years, Kopperberg county, Swedish Two-County Trial
| Primary therapy | Prophylactic |
| Quadrantectomy(QM) (18/203) | Radiotherapy (8/54) |
| Chemotherapy (1/1) | |
| Hormone (3/4) | |
| Chemo + hormone (1/1) | |
| None (5/143) | |
| Sub-mastectomy (SM) (2/5) | Radiotherapy (0/1) |
| Hormone (1/1) | |
| None (1/3) | |
| Total mastectomy without axillary dissection (TMNA) (13/46) | Radiotherapy (2/8) |
| None (11/38) | |
| Total mastectomy with axillary dissection (TMA) (164/693) | Radiotherapy (99/276) |
| Hormone (3/9) | |
| Chemotherapy (1/2) | |
| Radio + hormone (1/1) | |
| None (60/405) | |
| Radical mastectomy (RM) (7/17) | Radiotherapy (5/10) |
| None (2/7) | |
| Needle biopsy (2/4) | Radiotherapy (1/2) |
| None (1/2) | |
| None (11/13) | Chemotherapy (4/4) |
| Hormone (3/4) | |
| Chemo + hormone (3/3) | |
| None (1/2) | |
| Unknown (36/72) | Radiotherapy (0/1) |
| None (36/71) |
Data were stored at Uppsala University Data Centre. Multivariate survival analysis adjusting for tumour size, node status, histology type and detection mode was performed using the Cox regression model using the SAS PHREG procedure. For purposes of survival analysis, it was frequently necessary to combine categories in order to have enough data for stable survival probability estimates. For example, we frequently combined SM and TMNA into one category and TMA and RM into another. Also, histological type was categorized into three prognostic classes (10) for purposes of survival analysis, as follows:
Favourable: ductal carcinoma in situ, invasive ductal carcinoma of malignancy grade 1, mucinous carcinoma, tubular carcinoma;
Intermediate: invasive ductal carcinoma of malignancy grade 2, lobular carcinoma, medullary carcinoma;
Poor: ductal carcinoma of malignancy grade 3.
RESULTS
Table 1 shows numbers of breast cancers and deaths cross-tabulated by primary and prophylactic adjuvant therapy. For primary therapy, TMA and QM were the most common strategies. For prophylactic adjuvant therapy, radiotherapy and no adjuvant therapy were common compared with other adjuvant therapies. Fig. 1 and Table 1 show that women who underwent quadrantectomy (QM) seemed to have more favourable prognosis than women receiving mastectomy. Similarly, a poor prognosis was observed in those who received adjuvant therapy (Fig. 2).
Figure 1. Survival by types of surgery.
Figure 2. Survival by prophylatic adjuvant therapy, the Swedish Two-County Trial. HCR = any combination of hormone therapy and/or chemotherapy.
Clearly, this must be at least partly due to poorer prior prognosis of the patients selected for adjuvant therapy, particularly in terms of tumour size, node status, histological type and detection mode. Evaluation of the primary and adjuvant therapies in this study should therefore take these factors into consideration. This can be seen in Tables 2-4. The primary cancers with tumour size 21 mm or more (Table 2), lymph node invasion (Table 3) or histology with poor prognosis, i.e. ductal-grade 3 (Table 4), were more likely to be subject to TMA or the administration of chemotherapy, hormone therapy and radiotherapy than those with tumour size under 2 cm, node negative or histology with favourable prognosis. For detection mode (Table 5), clinically detected cancers such as interval cancers, control cancers before screen and refusers tended to receive mastectomy whereas screen-detected tumours received less radical treatment as they were more frequently small, node negative and of favourable histology type.
Table 2. Numbers of invasive breast cancer deaths/cases (%)* for primary and prophylactic adjuvant therapies cross-tabulated by tumour size in women aged 40-74 years, Kopperberg county, Swedish Two-County Trial
| Therapy | Tumour size (mm) | |||
| 1-10 | 11-15 | 16-20 | 21+ | |
| 1. Primary therapy | ||||
| Quadrantectomy (QM) | 5/83 (35%) | 3/57 (25%) | 3/17 (10%) | 6/12 (4%) |
| SM + TMNA[dagger] | 1/12 (5%) | 2/10 (4%) | 5/12 (7%) | 6/10 (3%) |
| TMA + RM[Dagger] | 9/133 (57%) | 23/149 (64%) | 29/135 (79%) | 110/256 (77%) |
| Others and unknown | 0/6 (3%) | 2/17 (7%) | 3/7 (4%) | 42/54 (16%) |
| Total | 15/234 | 30/233 | 40/171 | 164/332 |
| 2. Prophylactic | ||||
| Radiotherapy | 5/56 (24%) | 16/85 (36%) | 19/59 (35%) | 75/149 (45%) |
| Chemotherapy | - | - | - | 6/7 (2%) |
| Hormone therapy | - | 1/2 (1%) | 0/4 (2%) | 8/10 (3%) |
| Chemo + hormone | - | - | - | 4/4 (1%) |
| Radio + hormone | - | - | 1/1 (1%) | - |
| None | 10/178 (76%) | 13/146 (63%) | 20/107 (62%) | 71/162 (49%) |
| Total | 15/234 | 30/233 | 40/171 | 164/332 |
Table 3. Numbers of breast cancer deaths/cases (%)* for primary and prophylactic adjuvant therapies cross-tabulated by node status in women aged 40-74 years, Kopperberg county, Swedish Two-County Trial
| Therapy | Node status | ||
| Negative | Positive and distant metastases | Total | |
| 1.Types of surgery | |||
| Quadrantectomy (QM) | 11/139 (23%) | 6/12 (4%) | 17/151 |
| SM + TMNA | 4/17 (3%) | 2/4 (1%) | 6/21 |
| TMA + RM | 64/419 (69%) | 100/225 (79%) | 164/644 |
| Others and unknown | 6/29 (5%) | 36/45 (16%) | 42/74 |
| Total | 85/604 | 144/286 | 229/890 |
| 2. Prophylactic | |||
| Radiotherapy | 29/156 (26%) | 83/181 (63%) | 112/337 |
| Chemotherapy | - | 6/7 (2%) | 6/7 |
| Hormone therapy | 0/1 (0%) | 8/13 (5%) | 8/14 |
| Chemo + hormone | - | 4/4 (1%) | 4/4 |
| Radio + hormone | - | 1/1 (0%) | 1/1 |
| None | 56/447 (74%) | 42/80 (28%) | 98/527 |
| Total | 85/604 | 144/286 | 229/890 |
Table 4. Numbers of breast cancer deaths/cases (%)* for primary and prophylactic adjuvant therapies cross-tabulated by histological type in women aged 40-74 years, Kopperberg county, Swedish Two-County Trial
| Therapy | Histology | ||
| Favourable[dagger] | Intermediate | Poor | |
| 1. Types of surgery | |||
| Quadrantectomy (QM) | 5/83 (32%) | 7/83 (19%) | 6/30 (10%) |
| SM + TMNA | 2/14 (5%) | 5/20 (5%) | 7/13 (4%) |
| TMA + RM | 11/156 (60%) | 69/287 (68%) | 87/233 (76%) |
| Others and unknown | 0/8 (3%) | 12/34 (8%) | 22/29 (10%) |
| Total | 18/261 | 93/424 | 122/305 |
| 2. Prophylactic | |||
| Radiotherapy | 8/61 (23%) | 44/143 (34%) | 61/139 (46%) |
| Chemotherapy | - | 1/2 (0%) | 3/3 (1%) |
| Hormone therapy | 2/3 (1%) | 3/6 (2%) | 2/5 (2%) |
| Chemo + hormone | - | - | 1/1 (0%) |
| Radio + hormone | - | - | 1/1 (0%) |
| None | 8/197 (76%) | 45/273 (64%) | 54/156 (51%) |
| Total | 18/261 | 93/424 | 122/305 |
Table 5. Numbers of breast cancer deaths/cases (%)* for primary and prophylactic adjuvant therapies cross-tabulated by detection mode in women aged 40-74 years, Kopperberg county, Swedish Two-County Trial
| Therapy | Detection mode | ||||||
| First screen | Later screen | Interval cancers | Refuser | Control before screen | Control at first screen | Total | |
| 1. Primary therapy | |||||||
| Quadrantectomy (QM) | 0/13 (6%) | 7/104 (40%) | 4/22 (15%) | 2/7 (13%) | 3/20 (8%) | 1/35 (34%) | 17/201 |
| SM + TMNA | 2/6 (3%) | 3/14 (5%) | 1/6 (4%) | 1/2 (4%) | 5/12 (5%) | 1/9 (9%) | 13/49 |
| TMA + RM | 29/188 (90%) | 22/136 (53%) | 23/90 (63%) | 12/33 (59%) | 71/184 (72%) | 6/53 (51%) | 163/684 |
| Others and unknown | 1/1 (0%) | 1/3 (1%) | 7/24 (17%) | 14/14 (25%) | 24/39 (15%) | 2/7 (7%) | 49/88 |
| Total | 32/208 | 33/257 | 35/142 | 29/56 | 103/255 | 10/104 | 242/1022 |
| 2. Prophylactic | |||||||
| Radiotherapy | 17/143 (69%) | 12/173 (67%) | 14/77 (54%) | 19/38 (68%) | 46/147 (58%) | 4/78 (75%) | 112/656 |
| Chemotherapy | 15/65 (31%) | 21/84 (33%) | 16/60 (42%) | 7/14 (25%) | 47/95 (37%) | 5/22 (21%) | 111/340 |
| Hormone therapy | - | - | - | 3/3 (5%) | 3/3 (1%) | - | 6/6 |
| Chemo + hormone | - | - | 2/2 (1%) | 0/1 (2%) | 5/8 (3%) | 1/4 (4%) | 8/15 |
| Radio + hormone | - | - | 2/2 (1%) | - | 2/2 (1%) | - | 4/4 |
| None | - | - | 1/1 (1%) | - | - | - | 1/1 |
| Total | 32/208 | 33/257 | 35/142 | 29/56 | 103/255 | 10/104 | 242/1022 |
Tables 2 and 3 show that the decision whether to use TMA or QM depends on node status and tumour size and that the critical point leading to a decision is approximately at 15 mm (Table 2). Table 4 shows there is no clear pattern for treatment by histological type, perhaps partly because this was not known with certainty until after operation. Mastectomy with axillary dissection was applied to 47% of DCIS and lobular cancers. The proportion receiving TMA was approximately 65-75% for other tumours irrespective of tumour type. This raises the possibility of over-treatment for histological types with favourable prognosis. Similarly, unlike subsequent screens and the control group at first screen, 90% of breast cancers diagnosed at the first screen in the ASP were operated by TMA (Table 5). This also suggests over-treatment for some slow-growing tumours, which are more likely to be found at first screen.
Figures 3 and 4 show that the difference in survival between types of primary surgery can be largely accounted for by the size and node status of the primary tumour. When survival is stratified by primary therapy and histological type (Figure 5), no substantial differences are observed between categories of surgery for the favourable tumour types. For those with tumour types with poor prognosis, a difference can be seen, with greater survival associated with quadrantectomy (due to confounding with tumour size and node status), but interestingly better survival among mastectomies with axillary clearance than without.
A
 |
B
|
C
| D
|
Figure 3. Survival by tumour size and surgery.
A
 |
B
|
Figure 4. Survival by node status and surgery.
Figure 5. Survival by histological type and surgery.
There is also some indication of a survival benefit of more radical surgery in the tumours of size 20 mm or more. There is, however, scope for residual confounding, which may be adjusted for by multivariate survival analysis.
Fig. 6 shows survival by prophylactic adjuvant therapy and node status. Apart from the combination of chemotherapy and radiotherapy being associated with a poor prognosis (at that time only a small minority of tumours judged to be very dangerous received this), the differences in survival by adjuvant therapy can be largely accounted for by node status. Similar results were observed for tumour size and histological type (data not shown).
A
 |
B
|
Figure 6. Survival by node status and prophylatic adjuvant therapy.
Tables 6 and 7 show results of univariate and multivariate analysis by the Cox regression model, treating therapeutic adjuvant therapies as time-dependent variables. Without adjustment for tumour attributes and detection mode, primary therapy based on mastectomy seems more detrimental than conservative surgery (QM). Controlling for tumour attributes and detection mode, the prognosis associated with mastectomy such as RM and TMA is better, although not significantly so, than that associated with QM (Table 7).
Table 6. Analysis of the effect of primary and prophylactic adjuvant therapies on survival of breast cancer using the Cox regression model, adjusted for age at randomization and age at diagnosis, Kopperberg county, Swedish Two-County Trial
| Variable names | Hazard ratio | (95% CI) |
| 1. Primary | ||
| Quadrantectomy (QM) | 1.00 | - |
| Sub-mastectomy | 2.50 | (0.47-13.45) |
| Total mastectomy without axillary dissection (TMNA) | 1.72 | (0.67-4.42) |
| Total mastectomy with axillary dissection (TMA) | 1.23 | (0.58-2.59) |
| Radical mastectomy (RM) | 0.94 | (0.24-3.67) |
| None | 1.61 | (0.32-8.11) |
| Unknown | 1.52 | (0.61-3.83) |
| 2. Prophylactic adjuvant therapy | ||
| None | 1.00 | - |
| Radiotherapy | 1.14 | (0.78-1.69) |
| Chemotherapy | 3.46 | (0.64-18.65) |
| Hormone therapy | 0.97 | (0.33-2.83) |
Table 7. Multivariate analysis of survival by surgery, prophylactic adjuvant therapy, tumour size, node status, histology and detection mode using the Cox regression model, adjusted for age at randomization, age at diagnosis, further therapies and all other factors in the table, Kopperberg county, Swedish Two-County Trial
| Variable names | Hazard ratio | (95% CI) |
| 1. Primary therapy | ||
| Quadrantectomy (QM) | 1.00 | - |
| Sub-mastectomy (SM) | 1.63 | (0.25-10.75) |
| Total mastectomy without axillary dissection (TMNA) | 1.10 | (0.38-3.19) |
| Total mastectomy with axillary dissection (TMA) | 0.79 | (0.33-1.91) |
| Radical mastectomy (RM) | 0.43 | (0.10-1.88) |
| None | 2.59 | (0.46-14.64) |
| Unknown | 0.60 | (0.20-1.83) |
| 2. Prophylactic adjuvant therapy | ||
| None | 1.00 | - |
| Radiotherapy | 1.00 | (0.64-1.56) |
| Chemotherapy | 0.45 | (0.06-3.54) |
| Hormone therapy | 0.73 | (0.22-2.47) |
| 3. Histology | ||
| Favourable (ductal-grade 1, mucinous, tubular and DCIS) | 1.00 | - |
| Intermediate (ductal-grade 2, lobular and medullary) | 1.22 | (0.72-2.08) |
| Poor (ductal-grade 3) | 1.49 | (0.88-2.52) |
| 4. Tumour size (mm) | ||
| 1-9 | 1.00 | - |
| 10-14 | 1.54 | (0.59-4.00) |
| 15-19 | 1.51 | (0.59-3.86) |
| 20-29 | 2.23 | (0.92-5.38) |
| 30-49 | 2.51 | (0.98-6.41) |
| 50+ | 3.97 | (1.50-10.50) |
| 6. Node status | ||
| Node negative | 1.00 | - |
| Node positive | 0.92 | (0.59-1.43) |
| Distant metastases | 2.26 | (0.86-5.98) |
| 7. Detection mode | ||
| Control | 1.00 | - |
| Prevalent screen | 0.53 | (0.28-1.01) |
| Subsequent screen | 1.19 | (0.69-2.06) |
| Interval cancer | 0.86 | (0.50-1.49) |
| Refuser | 2.15 | (1.09-4.22) |
As regards adjuvant therapy, prophylactic radiotherapy is associated with the same survival as no prophylactic therapy regardless of adjustment for tumour attributes and detection mode (Tables 6 and 7). Prophylactic chemotherapy appears to be associated with poorer survival until it is adjusted for tumour attributes and detection mode, again indicating that the adjuvant therapy has been applied specifically in response to poorer prognostic tumour attributes.
After adjustment, primary therapy and prophylactic adjuvant therapy do not have significant effects on prognosis, in contrast to the very significant effects of tumour attributes and detection mode (Table 7).
DISCUSSION
Before discussing the above results, it is worth emphasizing that these data pertain to primary tumours treated in the late 1970s and early 1980s. One would not expect to see the same large numbers of total mastectomies now, although axillary dissection is still common for tumours of small size.
There are three main findings in this study. First, it can be concluded that tumour attributes, rather than primary and prophylactic adjuvant therapy, dominated the prognosis of breast cancer (see Table 7), partly because type of surgery and prophylactic therapy were determined by tumour attributes (Tables 2-4). Table 8 shows that the larger primary tumours and node-positive tumours are more likely to be operated by mastectomy. The survival analyses suggest that this practice is beneficial, particularly for larger tumours. For node-negative, small tumours, conservative breast cancer surgery appears to be sufficient. Table 9 shows that tumours with node involvement are also more likely to receive prophylactic adjuvant therapy. The possibility of over-treatment is raised by the observation that tubular and mucinuous carcinoma have a similar probability of radical treatment to lobular carcinoma and ductal carcinoma of grade 1 and 2. With the use of mammographic-pathological correlations and modern techniques of core biopsy, it is possible to know the histological type in some cases, in advance of open surgery.
Table 8. Odds ratios of receiving mastectomy (including SM, TMNA, TMA and RM) against QM by tumour size, node status and histology, Kopperberg county, Swedish Two-County Trial
| Variable names | Odds ratio | (95% CI) |
| 1. Tumour size (mm) | ||
| 1-10 | 1.00 | - |
| 11-15 | 1.67 | (1.07- 2.58) |
| 16-20 | 6.84 | (3.64-12.85) |
| 21+ | 13.94 | (6.81-28.53) |
| 2. Node status | ||
| Node negative | 1.00 | - |
| Node positive and distant metastases | 2.66 | (1.37-5.16) |
| 3. Histology | ||
| Good | 1.00 | - |
| Medium | 1.16 | (0.76-1.76) |
| Bad | 1.09 | (0.63-1.87) |
Table 9. Odds ratios of receiving prophylactic adjuvant therapy (including radiotherapy, chemotherapy and hormone therapy or other combinations) against no therapy by tumour size, node status and histology, Kopperberg county, Swedish Two-County Triall
| Variable names | Odds ratio | (95% CI) |
| 1. Tumour size (mm) | ||
| 1-10 | 1.00 | - |
| 11-15 | 1.73 | (1.12-2.66) |
| 16-20 | 1.43 | (0.87-2.33) |
| 21+ | 1.69 | (1.08-2.64) |
| 2. Node status | ||
| Node negative | 1.00 | - |
| Node positive and distant metastases | 6.97 | (4.91-9.89) |
| 3. Histology | ||
| Good | 1.00 | - |
| Medium | 1.09 | (0.74-1.61) |
| Bad | 1.44 | (0.94-2.20) |
It should also be noted that whatever the effect of treatment, the survival benefit associated with small tumours and node-negative tumours is considerably greater than any effect of treatment. This reinforces the findings of Duffyet al. (8) that tumour size, node status and malignancy grade have significant independent effects on survival. It also suggests that the optimum strategy for control of breast cancer is early detection while the cancer is small and localized, with efficient excision of the primary tumour.
The results raise concern about over-treatment for small cancers, in particular tumours of size 15 mm or less, node-negative cancers or histological types with good survival. Although the above suggests a reasonable policy of mastectomy and conservative surgery according to tumour size and node status, there were large numbers (close to two-thirds) of 1-10 mm tumours receiving mastectomy and one-fifth of these tumours received prophylactic radiotherapy (Table 2), despite the strongly suggestive evidence of the equivalent efficacy between QM and TM (Fig. 3) and between radiotherapy and no adjuvant therapy (Table 7). It is well known that tumours smaller than 11 mm in maximum diameter have a very good prognosis (10-year survival in excess of 90%), regardless of mode of detection, age or malignancy grade. This suggests first that in all ages 40-74 years, tumours diagnosed while small, whether by screening, symptomatic clinical diagnosis or breast self-examination in the interval between screens, are likely to be curable. Second, even aggressive, poorly differentiated cancers can be cured if detected sufficiently early.
It could be argued that the reason for using mastectomy for 1-10 mm breast cancers might be due to lymph node involvement. In the Swedish Two-County Trial, 7% of tumours of size 1-10 mm were node positive. This suggests that for most tumours of size less than 10 mm, axillary clearance is unnecessary. From Table 3, it can be seen that large numbers (again at least two-thirds) of node-negative tumours of all sizes were treated with mastectomy with axillary dissection. Considered another way, 64% of axillary dissections resulted in a node-negative finding.
Similar observations may be made for cancers of less aggressive histological type such as DCIS, tubular carcinoma and ductal-grade 1 (Table 9), Moreover, the survival associated with mastectomy and quadrantectomy for these tumours is almost the same. This suggests that primary treatment based on mastectomy for these cancers may be unnecessary.
We can use the Cox regression results to predict the prognosis of breast cancer allowing for tumour size, node status, histological type and primary and adjuvant therapy, constructing a prognostic index in the same manner as Spiegelhalter and Knill-Jones (13,14). This will be the subject of a separate paper.
It is relatively straightforward to calculate survival for other combinations of tumour attributes to provide important information for clinicians and patients to distinguish potential cases for less radical treatments. McGuire and Clark (15) have already defined risk groups using size, nuclear grade, estrogen and progesterone receptor status, DNA ploidy, S-phase, Cathid and Her-2. It may not be feasible to collect all this information in all cases. In practice, tumour size, node status, histological type and detection mode are likely to provide sufficient prognostic information.
The implications of the above findings are very important in the context of breast cancer screening. As screening yields a substantial proportion of node-negative, small cancers and histological types with favourable prognosis, it therefore not only reduces deaths from breast cancer, but also gives the opportunity to use breast-conserving surgery. For example, as shown in Tables 3 and 5, approximately two-thirds of breasts with small cancers or with ductal-grade 1, mucinous and tubular carcinoma could be treated with QM, possibly in conjunction with sentinel lymphadenectomy. Moreover, the ability to dispense with adjuvant therapy for these primary tumours with good tumour attributes due to early diagnosis by screening is a cost-effectiveness aspect of breast cancer screening.
The above results also suggest various hypotheses for future investigation. These include:
- Certain tumours are likely to be node negative or to obtain minimal, if any, benefit of axillary dissection. These include small tumours, notably of favourable histological type. A predictive index for node status in tumours diagnosed in the present era of mammography screening should be derived and validated. In cases of uncertainty, techniques such as sentinel lymphadenectomy can be employed.
- Adjuvant therapies immediately after surgery (prophylactic adjuvant therapy) do not appear to confer any benefit in terms of prognosis of small tumours.
- The advent of screening results in the diagnosis of many cases while they are still small and the mammography and subsequent assessment of suspicious cases yields a considerable amount of information on tumours before open surgery.
It is therefore worth committing resources to trials of the use of mammographic-pathological correlations and other information from the screening process to determine which tumours can and which cannot benefit from more radical surgery, axillary dissection and adjuvant therapies. It has long been acknowledged that among the benefits of mammographic screening is the increased number of tumours treatable with less radical therapy. The results above, together with those of other researchers (2-4), suggest that this is still not appreciated and that there is still considerable over-treatment. The necessity for such trials is therefore an urgent one.
References
Received May 19, 1999; accepted September 2, 1999
For reprints and all correspondence: Stephen W. Duffy, MRC Biostatistics Unit, Institute of Public Health, University Forvie Site, Robinson Way, Cambridge CB2 2SR, UK
Abbreviations: ASP, active study population; HCR, hormone therapy, chemotherapy and radiotherapy combination; PSP, passive study population; QM, quadrantectomy; RM, radical mastectomy; SM, sub-mastectomy; TMA, total mastectomy with axillary dissection; TMNA, total mastectomy without axillary dissection
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: jnl.info{at}oup.co.uk
Last modification:
Copyright© 1999 Foundation for the Promotion of Cancer Research.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  K. Armstrong, E. Moye, S. Williams, J. A. Berlin, and E. E. Reynolds Screening Mammography in Women 40 to 49 Years of Age: A Systematic Review for the American College of Physicians Ann Intern Med, April 3, 2007; 146(7): 516 - 526. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. B Miller Commentary: A defence of the Health Insurance Plan (HIP) study and the Canadian National Breast Screening Study (CNBSS) Int. J. Epidemiol., February 1, 2004; 33(1): 64 - 65. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||