Early breast cancer: predictors of breast recurrence for patients treated with conservative surgery and radiation therapy

Radiotherapy Elsevier RADION

and Oncology,

19 (1990) 29-41

29

00757

Early breast cancer: predictors of breast recurrence for patients treated with conservative surgery and radiation therapy * John Boyages 1,7,Abram Recht ‘, James L. Connolly2, Stuart J. Schnitt2, Rebecca Gelman 1,3, Hanne Kooy’, Susan Love4, Robert T. Osteen5, Blake Cady6, Barbara Silver’ and Jay R. Harris ’ ‘Joint Center for Radiation Therapy and Department of Radiation Therapy, Harvard Medical School (HMS), Boston, ‘Department of Pathology, Beth Israel Hospital and HMS, Boston, ‘Division of Biostatistics and Epidemiology, Dana-Farber and Harvard School of Public Health, Boston, 4Breast Unit, Faulkner Hospital, Boston, ‘Department

Cancer Institute

of Surgery, Brigham and

Women’s Hospital, HMS, Boston, 6Department of Surgery, New England Deaconess Hospital, HMS, Boston, Massachussetts,

and

‘39th Thomas Baker Fellow of the Royal Australasian College of Radiologists

(Received 28 July 1989, revision received 21 November

Key

words:

Breast cancer conservative

1989, accepted

surgery; Radiotherpay,

10 April 1990)

recurrence

Summary The identification of factors associated with breast recurrence following conservative surgery (CS) and radiation therapy (RT) is of potential use in refining patient selection criteria and treatment technique. In an attempt to define such factors we examined the relationship between various clinical, pathologic and treatment characteristics and the likelihood of breast recurrence in 783 patients with clinical stage I or II breast cancer treated between July 1968 and December 1982. Treatment consisted of complete gross excision of the primary tumor and RT to a total dose of at least 60 Gy to the primary site. During this period, pre-treatment mammograms and detailed histologic assessment of the margins of resection were not routinely performed. Median follow-up for surviving patients was 80 months. Thirteen patients (1.6%) were lost to follow-up. Ninety-one patients (12%) have developed a breast recurrence, corresponding to 5- and lo-year actuarial rates of 10 and 18 %, respectively. The major feature associated with breast recurrence was the presence of an “extensive intraductal component” (EIC + ). An EIC + tumor was seen in 28 y0 of evaluable cases with infiltrating ductal carcinoma and accounted for 60% of breast recurrences. Forty-three of 166 patients (26%) with EIC + tumors developed a breast recurrence

Address for correspondence

and reprint requests: Abram Recht, M.D., Joint Center for Radiation Therapy, 50 Binney St, Boston, MA 02115, U.S.A. * Presented in part at the 30th annual meeting of the American Society for Therapeutic Radiology and Oncology, New Orleans, October 1988.

0167-8140/90/$03.50

0 1990 Elsevier Science Publishers

B.V. (Biomedical

Division)

30 compared with 29 of 418 patients (7%) without an EIC (EIC - ) (p = 0.0001). The 5-year actuarial rates of breast relapse were 24 and 6%, respectively (p = 0.0001). Very young age (defined as 34 years of age or younger) was also a significant factor associated with the risk of breast recurrence. Very young patients comprised 8 y0 of the patient population and accounted for 16 y0 of breast recurrences. Fifteen of 6 1 very young patients (25 %) developed a breast recurrence compared with 76 of 722 older patients (11%) (p = 0.001). The corresponding 5-year actuarial rates of breast recurrence were 21 and 9% (p = 0.005). None ,of the other clinical or pathological factors examined by univariate analysis were significantly correlated with recurrence in the breast. A multivariate model of site of first failure (polychotomous logistic regression) also showed that EIC + tumors and very young age were the main factors associated with a high relative risk of breast recurrence. We conclude that EIC + tumors and very young age are associated with a high risk of breast recurrence for patients treated with limited excision prior to RT.

Introduction Conservative surgery and radiation therapy (CS + RT) has become the most common treatment for early stage breast cancer at many institutions. In order to refine patient selection criteria and the technique for this approach, it is useful to identify risk factors associated with breast recurrence following treatment. Such risk factors, however, have not been well established. In part, this may be due to the large variation in the specific implementation of the treatment from center to center. Risk factors associated with breast recurrence for patients treated with limited breast surgery may not apply to those treated with larger resections, such as a quadrantectomy. Differences in radiation dose, boost volume and the use and sequence of adjuvant chemotherapy may also explain why prognostic factors for breast recurrence vary between institutions. In previous reports, we found that among patients with infiltrating ductal carcinoma treated with a simple gross excision prior to radiation therapy, an extensive intraductal component in the tumor (EIC + tumor) was the most important factor associated with breast recurrence following CS + RT [ 2,11,23]. In this report we re-examine various patient, tumor and treatment characteristics as potential risk factors associated with a breast recurrence for a larger and updated group of patients with early breast cancer treated with CS + RT between 1968 and 1982. (Factors

related to the risk of distant failure are discussed elsewhere [7].) Our results continue to indicate that a high level of breast tumor control is achieved using this approach and that the main risk factor associated with a breast recurrence is the presence of an extensive intraductal component (EIC + ).

Materials and methods Nine hundred and twenty-eight women with UICC-AJCC clinical stage I or II breast cancer were treated at the Joint Center for Radiation Therapy (JCRT) between July, 1968 and December, 1982. To eliminate the confounding effects of possibly inadequate treatment, 31 patients who had less-than-excisional biopsy (i.e. needle biopsy or incisional biopsy only) and 114 patients who received a total dose of less than 60 Gy to the region of the primary tumor were excluded from the analysis. The study population thus consisted of 783 patients. In contrast to our previous studies, patients who presented with simultaneous bilateral tumors were excluded from this study and for patients with a metachronous opposite breast tumor, only the first tumor was considered in the analysis of breast recurrence. As of March 1988, median follow-up was 80 months for 600 surviving patients (range, 50-202 months). In addition, thirteen patients who were lost to follow-up without evidence of

31 disease were censored at intervals of between 17 and 122 months (median, 59 months). Median follow-up for all patients (dead, alive and LTFU) was 75 months (range, 4-202). Ninety-five percent of the 600 surviving patients had follow-up of at least 5 years. The median age at diagnosis was 52 years (range, 25-88). During this period, pre-treatment mammograms were not routinely performed. Surgery consisted of an excisional biopsy, defined as gross total removal of the primary tumor, usually without histologic confirmation of negative margins. Five hundred and ninety-seven patients (76 %) underwent axillary nodal dissections, usually limited to levels I and II. Details of the radiation treatment have been described elsewhere [ 10,20,26]. All patients received whole breast irradiation, typically to a total dose of 45-50 Gy over 5 weeks. Ten patients were treated in the early years of this study to whole breast doses of 60 Gy without a boost. A boost dose to the primary tumor site was given to 773 patients. The boost was delivered by an iridium-192 implant in 664 cases, electron beam in 79 and photon irradiation in six cases. Twenty-four patients had a mixed interstitial and external beam boost. Thirty percent of implants were one-plane and 70 % were two-plane or volume implants. A typical implant was 6-7 cm wide with a separation between planes of approximately 1.5 cm. The implant generally encompassed the entire surgical scar and its depth varied according to the original size and depth of the primary and on the size of the breast. Implants were performed using plastic coated platinum-filtered iridium- 192 wire which was divided into seeds 0.6 cm long, separated by 0.4 cm between seeds. The target volume minimum dose was usually specified as the isodose laterally between the outer-most strands of adjacent planes, when viewed end-on. The maximum dose accepted was generally less than about twice the minimum target dose. Dose rate was generally determined by implant geometry and seed strength relative to the target volume but was usually 40-50 cGy per hour. Implant volume was determined indirectly by using the number of

seeds per implant. This analysis, however, did have certain limitations as this method did not estimate distance between planes. Of note, however, larger single-plane implants and two-plane or volume implants were associated with a larger number of seeds. The total dose to the region of the primary tumor ranged from 60 to 84 Gy (median, 67 Gy). One hundred and forty-six patients (19%) received radiation to the breast only, 81 (10%) to the breast and supraclavicular fossa (SCF), and three to the breast and axilla. The remaining 553 patients (7 1%) were treated with a three-field technique to the breast, SCF and axilla. Chemotherapy was given to 187 patients (24%). For this study, we compared 164 patients (83 %) who received at least three cycles of either cyclophosphamide, methotrexate and 5-fluorouracil (CMF) or doxorubicin (Adriamycin) to 594 patients who received no adjuvant chemotherapy. Initial biopsy material was reviewed by two of the authors (J.L.C., S.J.S.) for 732 of the 783 breast tumors (94%). Pathologic assessment was made without knowledge of the patient’s clinical outcome. Details of the assesment have been previously described [23]. For this report, analysis of the impact of pathologic features was limited to the 584 of the 783 breasts (75%) in which the most aggressive histology was an infiltrating ductal carcinoma and sufficient breast tissue was present adjacent to the primary tumor for evaluation. Pathologic features examined included nodal status, estrogen receptor protein status (ERP), lymphatic and blood vessel invasion, nuclear and histologic grade of the tumor and the presence and extent of intraductal carcinoma within the infiltrating primary tumor and in adjacent grosslynormal breast tissue. An EIC + tumor was defined as the presence of intraductal carcinoma both within the primary infiltrating ductal tumor (comprising at least 25 y0 of the tumor area) and in adjacent grossly normal breast tissue, or as a predominantly intraductal tumor with one or more areas of focal invasion. A detailed histopathologic description of this entity will be the subject of a separate report.

32 Breast recurrence was defined as the detection of cancer in the treated breast or overlying skin occurring before or simultaneously with the diagnosis of distant metastases. (Eight patients developed breast recurrences after the diagnosis of distant metastases.) Breast recurrences were classified by location in the breast relative to the primary tumor site and boost volume as true recurrences (TR, within the area of the boost), marginal misses (MM, adjacent to the boost area), recurrences elsewhere (E, in other quadrants of the breast) and skin recurrences (S). For two patients, the type of recurrence could not be classified; for simplicity, these two patients have been classified with skin recurrences as “other”. The risk of breast recurrence was evaluated by the use of actuarial calculations and by crude incidence. Actuarial curves of any breast recurrence or of TR/MM alone were calculated by the method of Kaplan and Meier and comparisons between curves were made using the two-tailed “naive” log-rank test. The Kaplan-Meier estimates, however, can be shown to be unbiased and maximum likelihood estimates only if time to distant and time to breast failure are statistically independent. This assumption can be questioned in this data set. Such non-independent censoring could cause the log-rank test to be biased in either direction (too significant or not significant enough). Patients were censored from the calculation of breast recurrence at the time of last disease-free follow-up, the discovery of metastatic disease (153 patients), the discovery of regional nodal disease without a breast recurrence (7 patients), or death from intercurrent illness (29 patients). Patients were censored from the calculation of distant failure at the time of last follow-up, scoring failure at the time of first discovery of metastatic disease (isolated axillary nodal failures were excluded as an event). A multivariate analysis was performed using a polychotomous logistic regression of site of first failure (breast, distant or none) on clinical, pathologic and treatment variables. This analysis assumes that patients in various subgroups are at risk for similar lengths of follow-up time ; to satisfy

this assumption, 22 patients treated prior to 1976 were excluded from the model. Dummy variables were included in the model to control for characteristics which were not known in some patients (for example, pathologic nodal status, nuclear grade, histologic grade and EIC). Results are represented as relative risks. Each relative risk was calculated from the model, and hence is adjusted for the other prognostic factors in the model. The relative risk of the “breast vs. none” column (Table VII) is defined as a fraction with the numerator being the odds of a breast recurrence versus the odds of no recurrence in the presence of the associated prognostic and the denominator being the odds of a breast recurrence in the absence of the prognostic factor. The odds of breast recurrence vs. no recurrence is the probability of breast recurrence divided by the probability of no recurrence. The relative risk in the “breast vs. distant” column was similarly defined [ 81. Differences between proportions were tested by the chi-square test. P-values of 0.05 or less were considered statistically significant.

Results Univariate analysis With a median follow-up of 80 months, 91 of 783 patients (12%) have developed a breast recurrence. The corresponding 5- and lo-year actuarial rates of breast recurrence were 10 and 18% (Fig. 1). A recurrence in the region of the primary tumor (TR or MM) was the most common pattern of breast recurrence (Table I). The median time to breast recurrence differed by the pattern of breast recurrence; an E failure tended to occur later than a TR or MM. The median time to diagnosis was 48 months for TR, 35 months for MM and 58 months for failures outside the primary tumor region. The last TR or MM occurred 105 months after the commencement of radiation therapy and 80 % occurred within 5 years. In contrast, the last

33

I

%

I

20-1

1

oi

10

5 TIME

(YEARS)

Fig. 1. Breast recurrence: all patients treated with gross excision and at least 60 Gy to the primary site. The number of patients at risk at 5 and 10 years is shown in parentheses.

TABLE Pattern

I of breast recurrence.

Pattern

No. of LF

TR MM E Other

43 21 17 10

Total

91

Median (mths)

Range

41 23 19 11

48 35 58 22

14- 87 7-105 14-123 12- 96

100

31

7-123

% All failures

TR = true recurrence; MM = marginal miss; E = elsewhere; Other = unclassified (2) and skin recurrence (6); LF = local breast failure; No. = number.

young patients constituted 8 % of the study population and accounted for 16% of breast recurrences. Fifteen of 61 very young patients (25%) developed a breast recurrence compared with 76 of 722 older patients (11%) (p = 0.001). The corresponding 5- and lo-year actuarial breast recurrence rates were 2 1 and 56% for very young patients and 9 and 15% for older patients (p < 0.05) (Fig. 2). However, the confidence limits for the lo-year rates were large due to the small number of patients at risk at 10 years. T-stage, N-stage and pathologic nodal status were similar in the very young and older patients. We also assessed the incidence of distant recurrence for the two age cohorts and found no significant difference. The 5-year actuarial incidence of distant recurrence was 19% for patients aged 35 or older and 23 y0 for younger patients (p = 0.40). Table II shows the relationship between patient age and pattern of breast recurrence. In both cohorts recurrences occurred predominantly in the region of the primary tumor. Eighty percent of recurrences in very young patients and 68% of those in older patients were either a TR or an MM (p = NS).

Age

c.34:

5.yr.

----

%

Failure 21%(36) 9X(497) 0 005

IO-yr

56%(l)

15%(32) 0 001 _-

E failure occurred 123 months after the commencement of radiation therapy, and only 4 1y0 occurred within 5 years. We examined the relationship between various patient, tumor and treatment characteristics and the incidence and pattern of breast recurrence.

I

t

5

Patient characteristics We examined the relationship between patient age and the risk of breast recurrence and found that very young patients (< 34 years) had a higher incidence of recurrence than older patients. Very

Fig. 2. Breast recurrence: very young patients (age < 34) versus older patients (age > 35). All patients treated with gross excision and at least 60 Gy to the primary site. The number of patients at risk at 5 and 10 years is shown in parentheses.

34 TABLE Pattern

II of breast recurrence

related to age.

Age cohort (yrs)

No. of patients

No. of LF

TR (%)

MM (%)

E (%)

Other (%)

<34 >35

61 722

15 76

53 46

21 22

13 20

7 12

TR = true recurrence; MM = marginal miss; E = elsewhere; failure; No. = number.

Tumor characteristics The 5-year actuarial incidence of breast recurrence in relation to clinical and pathologic characteristics of the tumor is shown in Table III. Tumor location, clinical TN stage and UICC stage had no association with the incidence of breast recurrence. We examined the risk of a breast recurrence in relation to a variety of histopathologic features of the primary tumor. The main pathologic feature found to have a significant association with breast recurrence was an extensive intraductal component. Patients with EIC + tumors constituted 28 % of evaluable patients with infiltrating ductal carcinoma and accounted for 60% of breast recurrences in that group. Forty-three of 166 patients (26%) with EIC + tumors developed a breast recurrence compared with 29 of 418 patients (7%) with EICtumors (p = 0.0001). Patients with EIC + tumors had a 5-year actuarial probability of breast recurrence of 24%) compared with 6% for patients with EIC - tumors (p = 0.0001) (Fig. 3). The lo-year actuarial probability of a breast recurrence was 32 y0 for EIC + tumors and 14% for EIC - tumors (p = 0.0001). There was no significant difference in the 5-year actuarial incidence of distant recurrence for patients with EIC + tumors (14%) and for those with EIC - tumors (22%) (p = 0.4). The pattern of breast recurrence differed for EIC + and EIC - tumors (Table IV). Thirtyeight of 43 breast recurrences for EIC + tumors (88 %) occurred in the region of the primary tumor (TR or MM) compared with 16 of 29 breast recurrences for EIC - tumors (55 %) (p = 0.004). The

Other = unclassified

(2) and skin failures (6); LF = local breast

100

TIME

(YEARS)

Fig. 3. Breast recurrence: patients with EIC + tumors versus patients with EIC - tumors. Analysis limited to patients with infiltrating ductal carcinoma and evahtable specimens. All patients treated with gross excision and at least 60 Gy to the primary site. The number of patients at risk at 5 and 10 years is shown in parentheses.

5-year actuarial probability of a TR or MM was 21% for EIC + tumors and 4% for EIC - tumors (p = 0.0001). The lo-year actuarial probability of a TR or MM was 29 % for EIC + tumors and 5 y0 for EIC - tumors. Among patients with evaluable pathology, patients with EIC + tumors accounted for 70% of recurrences at the primary site. In order to assess the consistency of EIC as a risk factor for a breast recurrence, we examined its association with breast recurrence by year of treatment of the patient. We found that EIC + tumors had a significantly higher rate of breast recurrence than EIC - tumors for each timecohort examined (p < 0.05) (Table V). The 5-year

35 TABLE

III

Clinical and pathologic Factors

factors analyzed for impact on 5-year actuarial risk of breast recurrence. No. evaluable

% S-Year LF

P-value

61 122

21 9

0.005

461 41 281

11 9 9

421 362

10 10

NS

683 100

10

NS

389

10 10

NS

394

166 418

24 6

0.000 1

228 356

13

262 321

10 12

NS

119 464

9

NS

12

261 203 313

14 9

203 261

14 9

(A) Clinical Age: <34 >35 Tumor location: Lateral Central Medial T-stage: T, T, N-stage: N o/1,4 N Stag: I II

NS

1

(B) Pathologic EIC: + Highest nuclear grade: I/II III Mitoses: Scant/absent Moderate/marked Histologic grade: I/II III ERP status: Positive Negative Not known Nodal status: Negative Positive

9

9

LF = local breast failure; NS = not significant; EIC = extensive intraductal

actuarial risk of a TR or MM for patients EIC + tumors was approximately 20% for year cohort compared to 2-6x for patients EIC - tumors (all groups, p < 0.05). We examined the relationship between

with each with very

NS

NS

NS

component.

young age, the presence of an EIC + tumor and the risk of a breast recurrence. For patients with evaluable pathology, the incidence of EIC + tumors was slightly higher in the very young patients (32%) than in the older patients (28%)

36 TABLE Pattern

IV of breast recurrence

Intraductal component

as affected by an extensive intraductal

No. of breasts

166 418

EIC + EIC -

No. of LF

Pattern

MM (%)

E (%)

Other (%)

56 38

33 17

4 28

I 17

43 29

V

% 5-Year actuarial risk of breast recurrence related to an extensive intraductal component by year cohort oftreatment. Year

1968-1979 1980 1981 1982 Total

No. of EIC +

% LF

No. of EIC -

% LF

70 29 35 32

23 20 28 24

107 85 108 118

4 3 8 9

166

24

418

6

LF = local breast failure. All differences EIC + and EIC - tumors, p < 0.05.

(p = 0.65).

of breast failure (%)

TR (%)

TR = true recurrence; MM = marginal miss; E = elsewhere; failure; No. = number.

TABLE

component.

in LF between

Very young patients constituted 10% of the patients with EIC + tumors and 9% of the patients with EIC - tumors. The 5-year actuarial incidence of a breast recurrence was 37 y0 for the 17 patients in the < 34/EIC + subset and 23 % for the 149 patients in the >35/EIC + subset (p = 0.34). The 5-year actuarial incidence of a breast recurrence was 20% for 36 patients in the < 34/EIC - subset and 5 y0 for 382 patients in the > 35/EIC - subset (p = 0.001). The presence of lymphatic vessel invasion (LVI) in the breast resection specimen was also noted to be associated with an increased risk of breast recurrence but this was confounded by the presence of an EIC. One hundred and fifty-six patients had LVI + tumors and 428 had tumors that were negative or indeterminate for LVI (LVI - ). Patients with LVI + tumors constituted 27% of the study population and had a 17%

Other = unclassified

(2) and skin failures (6); LF = local breast

5-year actuarial rate of breast recurrence compared to 9% for other patients (p = 0.009). Of note, LVI was slightly more common for patients with EIC + tumors (3 1%) than for patients with EIC - tumors (25%) (p = 0.2). In order to study this association further, we examined the risk of breast recurrence in four patient subsets according to EIC and LVI status. For patients with EIC - tumors, the presence or absence of LVI did not affect the incidence of breast recurrence. The 5-year actuarial incidence of a breast recurrence was 9% for patients with EIC - /LVI + tumors and 5 y0 for patients with EIC - /LVI - tumors (p = NS). However, patients with EIC + /LVI + tumors had a signi&unly higher risk of breast recurrence (34%) than patients with EIC + /LVI - tumors (20%) (p = 0.001). Estrogen receptor protein (ERP) status had no significant association with the incidence of a breast recurrence. The 5-year actuarial rate of a breast recurrence was 9% for ERP-positive tumors compared with 14% for ERP-negative tumors (p = NS). Treatment characteristics

We also examined the influence of radiation treatment parameters on the risk of a breast recurrence. Evaluation of a dose effect was limited because all study patients received at least 60 Gy to the primary site of and because implant volume was indirectly measured using the number of 1921r seeds implanted. Total dose to the primary site and implant volume were not associated with the incidence of a breast recurrence (Table VI). In

37 particular, variations in total dose or implant volume to the primary tumor region did not have an impact on the incidence of a breast recurrence for patients with EIC + tumors. The 5-year actuarial incidence of a breast recurrence for patients with EIC + tumors treated to a dose of between 60 and 69.9 Gy was 22% compared with 36% for patients treated to a dose of 70 Gy or greater (p = NS). The 5-year actuarial incidence of a breast recurrence for patients with EIC + tumors treated with an implant volume of less than or equal to 50 seeds, 5 1 to 74 seeds or > 75 seeds was 25, 25 and 20%, respectively (all p = NS). We examined the influence of adjuvant chemotherapy on the risk of a breast recurrence (Table VI). The 5-year actuarial incidence of a breast recurrence was 7% for patients who received at least 3 cycles of CMF or Adriamycin chemotherapy compared with 11% for patients (p = 0.16). who did not receive chemotherapy The use of chemotherapy was associated with a lower breast recurrence rate among patients aged 49 years or younger. The 5-year actuarial incidence of a breast recurrence was 8% for patients

TABLE

who received chemotherapy compared to 17 9; for patients who received no adjuvant chemotherapy (p = 0.04). Only 56 patients 50 or older received chemotherapy; its use was not associated with a lower rate of breast recurrence. Multivariate analysis

Table VII displays the results of a polychotomous logistic regression relating site of first failure (breast, distant or none) to various prognostic factors. The factors associated with the highest relative risk of breast recurrence were EIC + tumors and age less than 34 years. The relative risk of a breast recurrence as opposed to no recurrence was 4.1 for patients with EIC + tumors compared with patients with EIC - tumors. The relative risk of a breast recurrence as opposed to no recurrence was 2.7 for patients aged 34 years or younger compared with older patients. The relative risk of breast recurrence as opposed to distant recurrence was 7.2 for patients with EIC + tumors compared to patients with EIC tumors. The relative risk of breast recurrence as

VI

5-Year actuarial risk of breast recurrence

as related to treatment

factors.

Factor

Total number

% 5-Year LF

P-value

Total dose to primary site (Gy) 60-69.9 >70 Implant volume (no. of seeds)

639 144

10 10

NS

224 226 225

12 11 8

NS

164 594

I 11

NS

108 250

8 17

0.04

56 344

6 6

NS

t50 51-74 > 75 Chemotherapy

(all patients)

yes no Age < 50 yrs yes no Age 3 50 yrs yes no LF = local breast recurrence.

38 TABLE

VII

Polychotomous first failure.

logistic regression:

relative risk of type of

Factor

Breast vs. none

Breast vs. distant

P-value

EIC + Age < 34 LVI+ HG-III Stage-T, PN+ HNG-III Adj CTX

4.1 2.1 1.4 1.4 1.1 1.1 1.0 0.5

1.2 2.0 1.0 0.7 0.7 0.7 0.9 0.5


LVI + = presence of lymphatic vessel invasion; HG = histologic grade; pN+ = pathologically positive nodes; HNG = highest nuclear grade; Adj CTX = use of adjuvant chemotherapy. Relative risk given is risk of failure with presence of factor versus risk of failure with absence of factor.

opposed to distant recurrence was 2.0 for patients aged 34 years or younger compared with older patients. For both of these prognostic variables (EIC + and age < 34) the relative risks in the breast/none and breast/distant columns are greater than 1.0, which indicates that the probability of breast recurrence is higher for patients with these features than patients without them. For the other two significant variables (stage T,, histologic grade III), the relative risk in the “breast vs. none” column is greater than 1.0 and the relative risk in the “breast vs. distant” column is less than 1.0. In these two subsets of patients, there were slightly more breast recurrences (compared to no recurrences) than expected and more distant recurrence (compared to breast recurrence) than expected.

Discussion In this report we have examined various patient, tumor and treament characteristics as potential risk factors for a breast recurrence in patients with stage I or II breast cancer treated with CS + RT. Our results in this updated and expanded series

continue to indicate a high level of breast tumor control, and that an “extensive intraductal component” was the main factor associated with a breast recurrence. EIC + tumors were found in 28 % of evaluable patients with infiltrating ductal carcinoma and accounted for 60% of breast recurrences. Very young patient age at diagnosis was also associated with an increased risk of breast recurrence. Very young patients comprised 8 % of the population and accounted for 16 % of breast recurrences. Prognostic factors for a breast recurrence following treatment with CS + RT vary in reports from different institutions. Bartelink et al. [ l] also reported that an EIC was the main risk factor for a breast recurrence among patients treated with a simple gross excision prior to radiation therapy. However, their 5-year actuarial rate of a breast recurrence for patients with EIC + tumors was 9 %, compared with 24 % in our series. It is possible that this dissimilarity reflects differences in treatment techniques and length of follow-up. Only 26 of 79 patients (33 %) with EIC + tumors were at risk at 5 years in Bartelink’s report, compared with 103 of 166 (62%) in our series. Investigators from the Institut Curie found, by univariate analysis, that EIC + tumors had a significantly increased rate of breast recurrence (p = 0.03), however, in contrast to our findings, an EIC + tumor was not a risk factor in a multivariate analysis [9]. An EIC + tumor has not been reported as an important risk factor for breast recurrence by other institutions [5,6,27]. We believe that this is most likely due to differences in the definition of EIC and in surgical techniques between these studies and ours. Patients in our series typically underwent a very limited gross excision of the tumor prior to radiotherapy. It is likely that with more extensive breast surgery and more careful attention to the margins of resection the importance of an EIC + tumor as a predictor of breast recurrence will be reduced. A high tumor grade has also been reported as a risk factor associated with breast recurrence by Clarke et al. [ 51. They found that 4% of patients with Bloom grade I or II tumors developed a breast recur-

39 rence compared with 12% of patients with grade III tumors (p = 0.001). In our series, tumor grade did not predict for a breast recurrence. Our data also confirm our previous observation that very young age (< 34 years) was also associated with a higher risk of breast recurrence [ 181. However, the very young group constituted only 8% of the entire population. Young age has also been found to be associated with an increased risk of breast recurrence by other investigators [4,6,13,15,16,27,28]. Forquet et al. [9] found young age, vascular invasion and an incomplete excision to be associated with an increased incidence of breast recurrence using a Cox multivariate regression model. Others, however, have found no increased risk of breast recurrence with young age [25]. The reasons for this association between very young age and breast recurrence are unknown, but perhaps may be linked to the patient’s hormonal status or perhaps a higher incidence of EIC + tumors in this group of patients. The implications of this finding are of concern but must to be interpreted in light of other factors such as the success rate of salvage therapy with breast recurrence after for patients CS + RT, the likelihood of local recurrence in very young patients following mastectomy; the influence of chemotherapy; and, ultimately, the relative survival rates of very young patients treated with mastectomy or with CS + RT. Results of mastectomy for breast recurrence following CS + RT have shown a high rate of salvage [ 191. Furthermore, Mathews et al. [ 141, found a similar incidence of breast or chest wall recurrence for very young patients (< 35 years) treated with CS + RT (13 %) or with mastectomy (12%). More research is required on the results of treatment with CS + RT or with mastectomy for very young patients in order to properly define treatment recommendations. We also found the use of chemotherapy and the absence of LVI to be associated with a lower incidence of breast recurrence in selected subgroups. The effect of chemotherapy was observed for premenopausal patients (~49 years) and is discussed in greater detail

elsewhere [ 221. The presence of LVI was associated with a higher rate of breast recurrence only in patients with EIC + tumors. For patients with EIC - tumors the incidence of breast recurrence was low, and independent of the presence of LVI. Previous studies from our institution have addressed possible reasons for the higher incidence of breast recurrence following treatment of EIC + tumors by a simple gross excision and radiation therapy. In a study of patients who underwent re-excision of the primary tumor site after an initial gross excision because of positive or very close margins of resection [ 241, we found that patients with EIC + tumors had a greater incidence of residual tumor in the re-excision specimen than did patients with EIC - tumors (88 vs. 48%, p = 0.002). For patients with an EIC, residual tumor was often widespread and was composed predominantly of intraductal carcinoma, whereas residual tumor in patients without an EIC usually consisted of only scattered microscopic foci of infiltrating and/or intraductal carcinoma. In a study of mastectomy specimens by Holland et al. [ 121 the extent of residual tumor in the breast after simulated gross excision of EIC + and EIC - tumors was assessed using a correlated pathologic-radiologic mapping technique. Patients with EIC + tumors were significantly more likely than patients with EIC tumors to have residual intraductal carcinoma in the remainder of the breast (71 vs. 28x, p < 0.00001). Of note, approximately 30% of patients with EIC + tumors had “prominent” residual intraductal carcinoma (6 or more low power fields) at least 2 cm beyond the edge of the primary tumor, compared with 2% of patients with EIC - tumors (p < 0.0001). This incidence of prominent residual carcinoma is almost identical to the lo-year incidence of a TR or MM for patients with EIC + and EIC - tumors, respectively, in the present study. Furthermore, in that study residual microscopic tumor was most often located in proximity to the primary tumor rather than randomly located throughout the remainder of the breast. This observation suggests that, for patients with EIC + tumors, a larger local resec-

40 tion would result in a lower residual microscopic tumor burden and might therefore be associated with a lower risk of a breast recurrence. The optimal approach to breast conservation for patients with stages I or II breast cancer requires a judicious combination of CS + RT. Either extensive breast surgery or very high dose radiation therapy can adversely affect the cosmetic result [ 3,171. Our studies indicate that combining local excision of the tumor and moderate doses of radiation therapy produces a good to excellent cosmetic result in the majority of patients [21] and is highly effective treatment for patients with EIC - tumors. For patients with EIC + tumors, limited breast surgery is associated with a high incidence of breast recurrence which is not affected by variations in boost dose and volume. For these patients, therefore, more extensive breast surgery may be required if CS + RT is to be employed. Our current practice is to determine the total extent of breast surgery based on the presence or absence of an EIC in the initial specimen and careful examination of excision margins. Patients typically first undergo a simple gross excision of the primary tumor with removal of sufficient adjacent breast tissue to assess the intraductal component. For patients with EIC - tumors (approximately 70% of evaluable patients), this simple gross excision is usually adequate surgical treatment. If the resection margins are prominently involved, a re-excision of the biopsy site is recommended, as we suspect that these may be the tumors that constitute most of the failures in the EIC - group. For patients with EIC + tumors, microscopic tumor involvement at the margins is common and we recommend a reexcision of the biopsy site to assess the presence and extent of any residual tumor. If the re-excision specimen shows marked residual carcinoma involving the new resection margins, then mastectomy is recommended. If the new margins of excision are clearly negative, then radiation therapy is considered to be a reasonable option. In order to assess the validity of this approach, we are currently performing an outcome study of

patients with EIC + tumors who underwent re-excision prior to radiation therapy.

a

Acknowledgements The authors would like to thank Lisa Sarta for data collection and management and for technical assistance. We also thank Beow Yeap, MA for statistical assistance.

References Bartelink, H., Borger, J. H., van Dongen, J. H. and Peterse, J. L. The impact of tumor size and histology on local control after breast-conserving therapy. Radiother. Oncol. 11: 297-303, 1988. Boyages, J., Recht, A., Connolly, J., Schnitt, S., Rose, M. A., Silver, B. and Harris, J. R. Factors associated with local recurrence as a first site of failure following the conservative treatment of early breast cancer. Recent Results. Cancer Res. 115: 92-102, 1989. Boyages, J., Barraclough, B., Middledorp, J., Gorman, D. and Langlands, A. 0. Early breast cancer - cosmetic and functional results after treatment by conservative techniques. Aust. NZ J. Surg. 58: 111-121, 1988. Calle, R., Vilcoq, J. R., Zafrani, B., Vielh, P. and Fourquet, A. Local control and survival ofbreast cancer treated by limited surgery followed by irradiation. Int. J. Radiat. Oncol. Biol. Phys. 12: 873-878, 1986. Clarke, D. H., Le, M. G., Sarrazin, D., Lancombe, M-J., Fontaine, F., Travagli, J-P., Francoise, M-L., Contesso, G. and Arriagada, R. Analysis of local-regional relapse in patients with early breast cancers treated by excision and radiotherapy. Experience of the Institut GustaveRoussy. Int. J. Radiat. Oncol. Biol. Phys. 11: 137-145, 1985. Delouche, G., Bachelot, F., Premont, M. and Kurtz, J.M. Conservation treatment of early breast cancer: long term results and complications. Int. J. Radiat. Oncol. Biol. Phys. 13: 29-34, 1987. Epstein, A. H., Connolly, J., Gelman, R., Schnitt, S., Silver, B., Boyages, J., Rose, M. A., Recht, A. and Harris, J. R. The predictors of distant relapse following conservative surgery and radiation therapy for early breast cancer are the same as those following mastectomy (Abstract). Int. J. Radiat. Oncol. Biol. Phys. 15 (Suppl. 1): 153, 1988. Fineberg, S. E. The Analysis of Cross-Classified Categorical Data. MIT Press, Cambridge, 1980. Forquet, A., Vilcoq, R., Zafrani, B., Durand, J. C. and

41

10

11

12

13

14

15

16

17

18

Mosseri, V. Early stage breast cancer: a multivariate analysis of the risk oflocal recurrence following conservative treatment. Long term results (Abstract). Int. J. Radiat. Oncol. Biol. Phys. 15 (Suppl. 1): 181, 1988. Harris, J. R., Botnick, L. E., Bloomer, W. D., Chaffey, J. T. and Hellman, S. Primary radiation therapy for early breast cancer: the experience at the Joint Center for Radiation Therapy. Int. J. Radiat. Oncol. Biol. Phys. 7: 1549-1552, 1981. Harris, J. R., Connolly, J. L., Schnitt, S., Cohen, R. and Hellman, S. Clinical-pathological study of early breast cancer treated by primary radiation therapy. J. Clin. Oncol. 1: 184-189, 1983. Holland, R., Connolly, J. L., Gelman, R., Mravunac, M., Hendriks, J., Verbeek, A., Schnitt, S. J., Recht, A., Silver, B., Boyages, J. and Harris, J. R. The presence of an extensive intraductal component in a limited excision specimen containing an infiltrating ductal carcinoma predicts for prominent residual disease in the remainder of the breast (Abstract). Int. J. Radiat. Oncol. Biol. Phys. 15: (Suppl. 1) 182, 1988. Kurtz, J. M., Brandone, H., Ayme, Y., Spitalier, J. M. and Amairic, R. Mammary recurrences in women younger than forty. Int. J. Radiat. Oncol. Biol. Phys. 14: 271-276, 1988. Mathews, R. H., McNeese, M. D., Montague, E. D. and Oswald, M. J. Prognostic implications of age in breast cancer patients treated with tumorectomy and irradiation or with mastectomy. Int. J. Radiat. Oncol. Biol. Phys. 14: 659-663, 1988. Muscolino, G., Luini, A., Bedini, A. V., Beretta, E. and Sacchini, V. Impact of young age on local recurrence risk in patients with early breast cancer undergone QU.A.RT. (Abstr.) Breast Cancer Res. Treat. 10: 105, 1988. Nobler, M. P. and Venet, L. Prognostic factors in patients undergoing curative irradiation for breast cancer. Int. J. Radiat. Oncol. Biol. Phys. 11: 1323-1331, 1985. Olivotto, I., Rose, M. A., Osteen, R. T., Love, S., Cady, B., Silver, B., Recht, A. and Harris, J. R. Late cosmetic outcome after conservative surgery and radiation therapy (Abstract). Int. J. Radiat. Oncol. Biol. Phys. 15 (Suppl. 1): 154, 1988. Recht, A., Connolly, J. L., Schnitt, S. J., Silver, B., Rose, M. A., Love, S. and Harris, J. R. The effect of young age on tumor recurrence in the treated breast after conservative surgery and radiotherapy. Int. J. Radiat. Oncol. Biol. Phys. 14: 8-10, 1988.

19 Recht, A., Schnitt, S. J. and Connolly, J. L., Rose, M. A., Silver, B., Come, S., Henderson, I. C., Slavin, S. and Harris, J. R. Prognosis following local or regional recurrence after conservative surgery and radiotherapy for early stage breast carcinoma. Int. J. Radiat. Oncol. Biol. Phys. 16: 3-9, 1989. 20 Recht, A., Silver, B., Schnitt, S., Connolly, J., Hellman, S. and Harris, J. R. Breast relapse following primary radiation therapy for early breast cancer. I. Classification, frequency and salvage. Int. J. Radiat. Oncol. Biol. Phys. 11: 1271-1276, 1985. 21 Rose, M. A., Olivotto, I., Cady, B., Koufman, C., Osteen, R., Silver, B., Recht, A. and Harris, J. R. The long-term cosmetic results of conservative surgery and radiation therapy for early breast cancer. Arch. Surg. 124: 153-157, 1989. 22 Rose, M. A., Henderson, I. C., Gelman, R., Boyages, J., Gore, S. M., Come, S., Silver, B., Recht, A., Connolly, J. L., Schnitt, S. J., Coleman, C. N. and Harris, J. R. Premenopausal breast cancer patients treated with conservative surgery, radiotherapy and adjuvant chemotherapy have a low risk of local failure. Int. J. Radiat. Oncol. Biol. Phys. 17: 711-717, 1989. 23 Schnitt, S. J., Connolly, J. L., Harris, J. R., Hellman, S. and Cohen, R. B. Pathologic predictors of early local recurrence in Stage I and II breast cancer treated by primary radiation therapy. Cancer 53: 1049-1057, 1984. 24 Schnitt, S. J., Connolly, J. L., Khettry, U., Mazoujian, G., Brenner, M., Silver, B., Recht, A., Beadle, G. and Harris, J. R. Pathologic findings on re-excision of the primary site in breast cancer patients considered for treatment by primary radiation therapy. Cancer 59: 675-681, 1987. 25 Solin, L. J., Fowble, B., Schultz, D. J. and Goodman, R. L. Age as a prognostic factor for patients treated with definitive irradiation for early stage breast cancer (Abstr). Int. J. Radiat. Oncol. Biol. Phys. 14: 184, 1987. 26 Svensson, G. K., Bjarngard, B. E., Larson, R. D. and Levene, M. B. A modified three-field technique for breast treatment. Int. J. Radiat. Oncol. Biol. Phys. 6: 689-694, 1980. 27 Van Limbergen, E., Van den Bogaert, W., van der Schueren, E. and Rijnders, A. Tumor excision and radiotherapy as primary treatment of breast cancer. Analysis of patient and treatment parameters and local control. Radiother. Oncol. 8: 1-9, 1987. 28 Vilcoq, J. R., Calle, R., Stacey, P. and Ghossein, N. A. The outcome of treatment by tumorectomy and radiotherapy of patients with operable breast cancer. Int. J. Radiat. Oncol. Biol. Phys. 7: 1327-1332, 1981.