**5. Recurrence rates in patients with small stellate breast cancers**

Survival in this group was generally excellent. Figure 1 shows survival by size group (1-9 mm and 10-14 mm) and radiotherapy. In all four groups, long-term survival was 90% or greater. There was no significant effect of radiotherapy on survival.

Fig. 1. Breast cancer specific survival of 425 1-14 mm stellate invasive breast cancers by size group and radiotherapy.

Table 1 shows the study subjects and recurrence rates by tumour, host, diagnostic and treatment features. There were 70 local recurrences in 425 patients. Overall, the average rate of recurrence was 1.6% per annum. Notably high recurrence rates were observed in grade 3 tumours and lobular carcinoma cases. Because 51% of the cases were diagnosed before 1990, the proportions of subjects treated with breast conserving surgery and with hormone therapy were considerably smaller than would be the case in tumours diagnosed at the present time.

Radiotherapy After Surgery for Small Breast Cancers of Stellate Appearance 221

55+ 2.53 0.93-6.87

BCS 1.81 0.93-3.51

Lobular 0.92 0.31-2.76

Yes 1.21 0.40-3.62

1977-84 1.00 - 1985-89 1.19 0.55-2.53 1990-94 2.15 0.95-4.82 1995-98 1.42 0.55-3.04

Factor Category Cox regression RR\* 95% CI Age group <55 1.00 -

Surgery Mastectomy 1.00 -

Histology Ductal, other 1.00 -

than age 55, no substantial benefit of radiotherapy was observed.

cases, being smaller than 15 mm, would receive wide local excision.

Grade Trend 1.93 1.31-2.85 Radiotherapy No 1.00 -

Interaction Lobular AND radiotherapy 3.58 0.90-14.13 Interaction Age 55+ AND radiotherapy 0.26 0.07-0.88

Table 2. Relative hazards and 95% confidence intervals from the final Cox regression model

Table 3 shows the results from the final Cox regression model, in terms of estimated annual rates of recurrence in the absence of radiotherapy and absolute reductions in cumulative 15 year recurrence associated with radiotherapy. Overall, absolute effects were small. For example, for grade 1, non-lobular tumours, in women aged 55 or older, treated with mastectomy, the absolute reduction in risk of local recurrence associated with radiotherapy was 1.87%. Substantial reductions in recurrence rates were observed only for cases aged 55 or more at diagnosis, of grade 2 or 3 and of non-lobular histological type. For some combinations of factors, notably lobular carcinoma cases younger than age 55, the reductions were negative, i.e. showing increases in recurrence in those treated with radiotherapy. For lobular carcinoma cases age 55 or over, and for non-lobular cases younger

This study pertains to a very special subgroup of good prognosis tumours, i.e. stellate lesions less than 15 mm in size without calcifications. Within this group, our results suggest that radiotherapy was substantially beneficial only in terms of preventing local recurrence in women aged 55 and older diagnosed with non-lobular carcinoma grade 2 or 3, a subgroup constituting 31% of the subjects (131 out of 425). Of the remaining 294 cases, 183 (63%) received radiotherapy, and based on these results not only did not substantially benefit from it, but may indeed have had their risk of recurrence increased. It might be argued that in modern treatment practice, very few of the mastectomy cases would receive radiotherapy. Where margins are close, the NCCN recommends that radiotherapy should be considered (NCCN 2009). However, also in modern therapeutic practice, a large proportion of these

Radiotherapy is primarily aimed at reducing the risk of local recurrence (Liljegren 2002). It has been suggested that radiotherapy might be dispensed with for low-risk patients (e.g.

Epoch of diagnosis

\* RR=rate ratio

after stepwise regression.

**6. Expert commentary** 


\* BCS=breast conserving surgery

Table 1. Host, tumour and treatment characteristics, with the corresponding rates of local recurrence.

Table 2 shows the final model from the stepwise Cox regression for prediction of recurrence. Age and epoch of diagnosis were included regardless of significance. A highly significant increase in risk of recurrence was noted with grade (HR=1.93, 95% CI 1.31-2.85, p=0.001, trend test). Although the main effect of radiotherapy was not significant, it had a significant interaction with age (p=0.03), associated with lower rates of recurrence in patients aged 55 and over (HR=0.26, 95% CI 0.07-0.88), but not in patients aged less than 55. It also had a borderline significant interaction with histological type (p=0.07) with a high rate of recurrence in lobular carcinoma cases treated with radiotherapy (HR=3.58, 95%CI 0.90-14.13). The effects of BCS (p=0.01) and histological type (p=0.04) were significant before adjustment for these interactions, so these factors were retained in the model. No other variables had significant effects on risk of recurrence after adjustment for the factors in Table 2.


\* RR=rate ratio

220 Imaging of the Breast – Technical Aspects and Clinical Implication

1977-84 117 (28) 20 1.4 1985-89 99 (23) 18 1.5 1990-94 99 (23) 19 2.0 1995-98 110 (26) 3 1.7

55+ 283 (67) 44 1.6

1-9 156 (37) 25 1.6 10-14 269 (63) 45 1.6

Positive 45 (11) 10 2.0 Not examined 39 (9) 5 1.7

2 173 (42) 31 1.8 3 34 (8) 10 2.9 Unknown 14 5 4.1

BCS\* 289 (68) 49 1.7

Yes 254 (60) 38 1.5 Unknown 3 1 3.3

Yes 12 (3) 3 2.9 Unknown 2 0 0.0

Yes 32 (8) 6 2.4 Unknown 2 0 0.0

Lobular 52 (12) 14 2.8 Tubular 80 (19) 7 0.8 Other 7 (2) 2 2.9

Screening 323 (76) 48 1.4

Table 1. Host, tumour and treatment characteristics, with the corresponding rates of local

Table 2 shows the final model from the stepwise Cox regression for prediction of recurrence. Age and epoch of diagnosis were included regardless of significance. A highly significant increase in risk of recurrence was noted with grade (HR=1.93, 95% CI 1.31-2.85, p=0.001, trend test). Although the main effect of radiotherapy was not significant, it had a significant interaction with age (p=0.03), associated with lower rates of recurrence in patients aged 55 and over (HR=0.26, 95% CI 0.07-0.88), but not in patients aged less than 55. It also had a borderline significant interaction with histological type (p=0.07) with a high rate of recurrence in lobular carcinoma cases treated with radiotherapy (HR=3.58, 95%CI 0.90-14.13). The effects of BCS (p=0.01) and histological type (p=0.04) were significant before adjustment for these interactions, so these factors were retained in the model. No other variables had significant

recurrence rate

Factor Category Cases (%) Recurrences Crude annual %

Age group <55 142 (33) 26 1.5

Node status Negative 341 (80) 55 1.5

Grade 1 204 (50) 24 1.1

Surgery Mastectomy 136 (32) 21 1.3

Radiotherapy No 168 (40) 31 1.7

Chemotherapy No 411 (97) 67 1.6

Hormone therapy No 391 (92) 64 1.5

Histological type Ductal 286 (67) 47 1.6

Detection mode Symptomatic 102 (24) 22 2.0

effects on risk of recurrence after adjustment for the factors in Table 2.

Epoch of diagnosis

Tumour size (mm)

\* BCS=breast conserving surgery

recurrence.

Table 2. Relative hazards and 95% confidence intervals from the final Cox regression model after stepwise regression.

Table 3 shows the results from the final Cox regression model, in terms of estimated annual rates of recurrence in the absence of radiotherapy and absolute reductions in cumulative 15 year recurrence associated with radiotherapy. Overall, absolute effects were small. For example, for grade 1, non-lobular tumours, in women aged 55 or older, treated with mastectomy, the absolute reduction in risk of local recurrence associated with radiotherapy was 1.87%. Substantial reductions in recurrence rates were observed only for cases aged 55 or more at diagnosis, of grade 2 or 3 and of non-lobular histological type. For some combinations of factors, notably lobular carcinoma cases younger than age 55, the reductions were negative, i.e. showing increases in recurrence in those treated with radiotherapy. For lobular carcinoma cases age 55 or over, and for non-lobular cases younger than age 55, no substantial benefit of radiotherapy was observed.

## **6. Expert commentary**

This study pertains to a very special subgroup of good prognosis tumours, i.e. stellate lesions less than 15 mm in size without calcifications. Within this group, our results suggest that radiotherapy was substantially beneficial only in terms of preventing local recurrence in women aged 55 and older diagnosed with non-lobular carcinoma grade 2 or 3, a subgroup constituting 31% of the subjects (131 out of 425). Of the remaining 294 cases, 183 (63%) received radiotherapy, and based on these results not only did not substantially benefit from it, but may indeed have had their risk of recurrence increased. It might be argued that in modern treatment practice, very few of the mastectomy cases would receive radiotherapy. Where margins are close, the NCCN recommends that radiotherapy should be considered (NCCN 2009). However, also in modern therapeutic practice, a large proportion of these cases, being smaller than 15 mm, would receive wide local excision.

Radiotherapy is primarily aimed at reducing the risk of local recurrence (Liljegren 2002). It has been suggested that radiotherapy might be dispensed with for low-risk patients (e.g.

Radiotherapy After Surgery for Small Breast Cancers of Stellate Appearance 223

The results of this observational study suggest that contrary to standard practice, postoperative radiotherapy may not be the ideal treatment for all breast cancers treated with breast conserving surgery, particularly those with good prognosis. This is not to deny the results of the randomised trials and meta-analyses. There is clear evidence from these that radiotherapy reduces local recurrence and improves survival. However, this does not necessarily imply that it is needed in all cases. There is potential for utilising patient and tumour information to assign treatment based on that which is appropriate for the subgroup. This tailored therapeutic approach uses simply obtained specifics, e.g patient age, radiological appearance and tumour histology/grade. It would enable a more accurate riskbenefit analysis to be calculated before prescription of therapies with adverse side-effects such as radiotherapy. It is therefore attractive in comparison with universal provision of radiotherapy to all patients. Before such policies can be implemented, it is essential that we are certain of the risks:benefit ratio for each patient subgroup and therefore these findings must be validated. This approach of investigating the level of benefit to different patient populations may be useful for other cancer therapies with adverse side-effects, with the objective of identifying other areas for improvement as medical oncology progresses to an

Radiotherapy is widely used to reduce the risk of local recurrence of breast cancer, particularly after breast conserving surgery. However, radiotherapy to the breast has adverse long-term side effects (risk of heart disease, lung cancer, angiosarcoma, deformation), and therefore it would be useful to identify subsets of patients for whom this treatment is unnecessary. Patients with stellate tumours of 1-14mm have a good prognosis and a high proportion of them might benefit from omitting radiotherapy. A Cox regression was applied to follow up data from 425 such patients and a comparison of local recurrence rates made for different patient groups/tumour stages receiving or not receiving radiotherapy post surgery. These observations suggest that the only group of patients within the 1-14 mm stellate lesions to benefit from radiotherapy are those aged 55 or more, with high grade (2 or 3) disease and non-lobular histology. Radiotherapy may not be beneficial to certain groups with higher risk of recurrence (e.g. younger women or lobular carcinoma) and some groups with low risk of recurrence (e.g. low grade tumours). Further

validation using subgroup analyses of trials already performed would be useful.

the American Cancer Society through a gift from the Longaberger Company.

*American journal of roentgenology* 187 (1):29-37.

Fieldwork for this study was supported by Cancer Research UK. Analysis was supported by

Alexander, M. C., B. C. Yankaskas, and K. W. Biesemier. (2006). Association of stellate

mammographic pattern with survival in small invasive breast tumors. *AJR.* 

**7. Five year view** 

era of individually tailored treatments.

**8. Conclusion** 

**9. Acknowledgements** 

**10. References** 


Table 3. Estimated recurrence rates in the absence of radiotherapy and absolute reduction in 15-year risk of recurrence by surgery, age, histological type and grade of tumours, from the final Cox regression model.

older patients with lower stage non-lobular tumours) (Liljegren 2002; Liljegren et al. 1997). One study has found risk of recurrence to be particularly low in those with non-dense breast tissue (Cil et al. 2009). The results here suggest that a significant proportion of patients with small stellate lesions can be considered at low risk, and that some higher risk patients, such as lobular carcinoma cases, may have high local recurrence rates despite radiotherapy. These results are observational, and need to be validated.

Another observational study has found, contrary to our results, that radiotherapy is associated with substantially reduced risk of local recurrence in lobular carcinoma (Diepenmaat et al. 2009). There was, however, a comparatively shorter follow up time (median of 7.2 years). Issues such as this may be resolved by delineating the tumour populations in radiotherapy trials which have already been conducted, or by carrying out new prospective trials.

The point has already been made that small stellate lesions are a good candidate for less aggressive therapy (Smith et al. 2004; Alexander, Yankaskas, and Biesemier 2006). The potential to save almost 70% of patients in this group from the hazards of radiation therapy is a goal worth pursuing.
