Journal of Surgical Oncology 2001;78:2±7

Do Surgical Oncologists Achieve Lower Rates ofLocal-Regional Recurrence in Node Positive Breast

Cancer Treated With Mastectomy Alone?

STEVEN LATOSINSKY, MD, MSc1

AND

HARRY D. BEAR, MD, PhD2*

1Assistant Professor, Department of Surgery, Division of Surgical Oncology, University of Manitoba,Winnipeg, Manitoba, Canada

2Professor and Chairman, Division of Surgical Oncology, The Medical College of Virginia, Virginia CommonwealthUniversity, Richmond, Virginia

Background and Objective: Adjuvant radiotherapy for node positivebreast cancer postmastectomy has been recommended by two previouslypublished randomized controlled trials (RCT). The local-regionalrecurrence rates in the control arms, however, were considered by somecritics to be excessive (>25% at 10 years). Inadequate surgery, asevidenced by the low number of axillary nodes reported, may have resultedin the high local-regional recurrence rates, allowing for the bene®ts seenwith radiotherapy. Fellowship trained surgical oncologists might provide`̀ better quality'' surgery, resulting in lower recurrence rates and thusmaking adjuvant radiotherapy unnecessary. Our objective was to establishthe local-regional control rate postmastectomy in node positive breastcancer patients operated on by surgical oncologists, and to determine iftreatment recommendations from previous RCTs are generalizable.Methods: Node positive stage IIb and IIIa breast cancer patients treatedwith mastectomy at the Medical College of Virginia Hospitals by surgicaloncologists, without adjuvant radiotherapy, and entered into adjuvantchemotherapy trials between 1978 and 1993 were identi®ed retro-spectively. Pathology and follow-up records were reviewed.Results: One hundred and thirty-seven patients were identi®ed. A medianof 18 axillary nodes was reported with a median of 4 positive nodes. Thelocoregional recurrence at 10-years was 27% (95% con®dence interval,19±35%).Conclusion: Despite some evidence of `̀ better quality'' surgery, there wasno clinically signi®cant difference in the local-regional recurrence rate inthis case series compared to controls in two previous RCTs. Recommen-dations for postmastectomy radiotherapy should be considered for nodepositive breast cancers, even if operated upon by surgical oncologists.J. Surg. Oncol. 2001;78:2±7. ß 2001 Wiley-Liss, Inc.

KEY WORDS: breast neoplasms; lymphatic metastasis; adjuvant radiother-apy; surgery; mastectomy; modi®ed radical; outcomes assessment

INTRODUCTION

In 1997, two randomized trials showing that post-mastectomy radiation therapy decreased local-regionalrecurrence rates and improved survival followingmastectomy in high-risk breast cancer patients were

*Correspondence to: Harry D. Bear, MD, PhD, PO Box 980011, 1200 E.Broad Street, 7th Floor West Wing, Richmond, VA 23298-0011.E-mail: hbear@hsc.vcu.edu

Accepted 7 June 2001

ß 2001 Wiley-Liss, Inc.

published [1,2]. Many surgeons were reluctant to acceptthe recommendations based on these trials due to con-cerns with the quality of surgery received by patients inthe trials. They felt the surgery in the Danish trial [1] wasof `̀ poor quality,'' re¯ected by the reporting of a medianof only seven nodes in the axillary dissection specimens.Poor quality surgery could have resulted in what wasperceived to be an excessive local-regional recurrencerate of 32% at 10 years in the control arm of the Danishtrial and allowed for the improvements seen in theradiotherapy group in local-regional recurrence andsurvival. Although, the Canadian trial [2] appeared tohave `̀ better quality'' surgery, as re¯ected in a median ofeleven axillary nodes reported, the local-regional recur-rence of 25% at 10 years postmastectomy was still felt tobe excessive. Mastectomies performed by fellowshiptrained surgical oncologists or by surgeons with a specialinterest in breast cancer might be more consistently of`̀ better quality'' and thus result in better outcomes,particularly local-regional recurrence.

A retrospective review of node positive mastectomypatients at the Medical College of Virginia Hospitals(MCVH) was undertaken to establish our own local-regional recurrence rates. All surgeries were performedby surgical oncologists. We wanted to determine if therecommendation for postmastectomy radiotherapy innode positive patients, based on recent trials, wasgeneralizable to our own patient population.

MATERIALS AND METHODS

Clinical Data

A retrospective case series of node positive breastcancer patients with mastectomy treated by fellowshiptrained surgical oncologists was conducted from recordsat MCVH. Patients with skin or chest wall involvementor having adjuvant radiotherapy to the chest wall wereexcluded. In order to have prospectively collected andcomplete follow-up data, subjects were restricted topatients who entered into adjuvant chemotherapy trialsbetween 1978 and 1993. Charts of eligible patients werereviewed to obtain the age at the time of entry into thetrials and pathology of the presenting breast cancerincluding size of the primary, the number of nodesexcised, and the number of positive nodes identi®ed.Biannual follow-up data was examined to determinethe date and location of ®rst local-regional recurrencefollowing mastectomy and/or the date of last follow-upor date of death. Local-regional recurrence was de®nedas ipsilateral recurrence of breast cancer, reportedon follow-up in one or more of the following sites:chest wall, internal mammary nodes, axillary nodes,infraclavicular nodes, and/or supraclavicular nodes,

consistent with the de®nitions of the Danish and Cana-dian trials.

Analysis

All statistical analyses were run on Statistical Packagefor the Social Sciences (SPSS) version 8.0.

Validity

To examine some of the potential biases in this study,variations in the local-regional recurrence rate over theperiod studied and the overall survival of women in thisstudy were examined. The former was performed bydividing consecutive cases into four equal sized groups,determining the local-regional recurrence rate for eachgroup and looking for temporal trends in the data.Statistical signi®cance was determined using a �2-test.The latter was performed to ensure that survival by stagein this group of patients was reasonable. This involvedgenerating Kaplan-Meier survival curves by stage forpatients who entered into the study and comparing it topublished survival data.

Local-regional Recurrence

The main outcome measure was the local-regionalrecurrence at 10 years estimated by the Kaplan-Meiermethod for the entire study group and then only forpatients with stage IIb disease. The former group wassimilar to those patients who entered into the Danishstudy while the latter were similar to patients in theCanadian study. The pattern of local-regional recurrencefollowing mastectomy in node positive disease wasstudied by tabulating the frequency of failure by site.

Univariate Analysis

As a secondary analysis, baseline demographic andpathological variables were examined individually tolook for predictors of local-regional recurrence. Variableswere categorized based on the important clinical criteria,age less than or greater than 55 years as a surrogate ofpre-vs. postmenopausal, tumor size categories basedon the American Joint Commission on Cancer (AJCC)breast cancer staging system [3], greater than or less thanor equal to three positive nodes based on considerationsdiscussed by Marks and Prosnitz [4] and the numberof nodes removed (categorized into intervals of 10).Kaplan±Meier survival curves for the variable of interestwere generated and statistical signi®cance was deter-mined using a log rank test.

Multivariate Analysis

A Cox proportional hazards model was used todetermine whether there were signi®cant predictors for

Node Positive Postmastectomy Recurrence 3

local-regional recurrence by multivariate analysis. Thesame variables as used in the univariate analysis wereentered into the multivariate analysis. A backwardelimination was used with a likelihood ratio level ofsigni®cance of 0.05 to reject and a forward acceptanceof 0.05.

RESULTS

Patient Characteristics

A total of 140 patients were identi®ed. Three patientswere excluded from the analysis due to incompleterecords leaving 137 patients for the analysis. No patientswere lost to follow-up with a mean and median follow-upof 9.8 and 7.9 years, respectively. Patient age, pathology,and stage distribution are characterized in Table I.All of these women received some form of systemictherapy. Approximately half received melphalan and 5-¯uorouracil alone or in combination with another agent.Approximately a third received AC or CMF alone or incombination with another agent. Approximately a third ofwomen were placed on tamoxifen, for the most part incombination with chemotherapy.

Validity

Figure 1 illustrates that cases examined are relativelyevenly distributed over the years of the study, as was theyear of entry of patients who developed a local-regionalrecurrence. Groupings based on the date of initialtreatment were not signi®cantly different in local-regional recurrence based on a �2-test.

The Kaplan±Meier survival curves by stage of diseaseare given for patients in this series in Figure 2. Survival at5 years of 66% (95% con®dence interval 56±76%) forstage IIb patients and 61% (95% con®dence interval56.9±75.1%) for stage IIIa patients are reasonablyconsistent with the respective survivals of 68 and 51%at 5 years based on data taken from the National CancerData Base and reported in the AJCC staging manual [3].

Local-Regional Recurrence

There were 34 local-regional recurrences. Table IIsummarizes the sites and frequencies of local-regionalrecurrences. The chest wall was the most frequent site oflocal-regional recurrence, with involvement in over 60%of cases.

There was a 27% (95% con®dence interval 18±36%)local-regional recurrence rate at 10 years in all patients,while it was 26% (95% con®dence interval 16±36%) inpatients with stage IIb, disease as seen from Figure 3.

Univariate Analysis

In the univariate analysis looking at age, primary tumorsize, number of nodes removed from axilla, and thenumber of nodes positive, no signi®cant differences couldbe shown between categories and the risk of local-regional recurrence, although our sample size was small.

Multivariate Analysis

There were no variables identi®ed as predictors oflocal-regional recurrence by multivariate analysis, butagain the sample size was small.

DISCUSSION

The surgical factors that may in¯uence outcomes in-clude specialty training or focus, case volume, or researchfocus [5]. The surgeon's role in outcomes has been exam-ined in various areas of oncology including colorectal[6,7], pancreas [8], and breast cancer [9]. In this study, itwas hypothesized that surgeons with specialized trainingin surgical oncology and interest in breast cancer mightprovide `̀ better quality'' surgery. This would, in turn, bepredicted to low result in local-regional recurrence ratesfollowing mastectomy in node positive women thatwould be comparable to the adjuvant radiotherapy armsin the Danish and Canadian trials and thus preclude theneed for adjuvant radiotherapy. In these trials the local-regional recurrence rates were 9 and 13% in the adjuvant

TABLE I. Descriptive Statistics of Patients Treated With Mastectomy for Node Positive BreastCancer at MCVH

Variable No. of observations Mean Median Standard deviation

Age (years) 136 49.6 49 11.1Primary size (cm) 134 4.2 4 2.2Number of axillary nodes

removed 137 19.2 18 8.6Number of positive nodes 137 6.6 4 6.93Stage

IIb 88IIIa 46Unknown (size of 3

primary uncertain)

4 Latosinsky and Bear

radiotherapy arms and 32 and 25% in the control arms,respectively, at 10 years.

The local-regional recurrence rates for node positivemastectomy patients in this study and other studiestreated without adjuvant radiotherapy are summarized inTable III. For all patients included in this study, theoverall local-regional recurrence at MCVH was 27% at10 years. These patients are comparable to control pat-ients in the Danish trial who had a 32% local-regionalrecurrence at 10 years. When the subset of MCVH pat-ients with node positive stage IIb disease are examined,the local-regional recurrence rate at 10 years was 26%.These patients are comparable to control patients in theCanadian trial who had a 25% local-regional recurrencerate at 10 years. Thus, local-regional recurrence at MCVHis similar to the controls in the Danish and Canadian trialsdespite all procedures being performed by fellowshiptrained surgical oncologists. The surgeons in the Danishand Canadian trials were not characterized, but werelikely of more diverse backgrounds and interests.

The Eastern Cooperative Oncology Group (ECOG)retrospectively reviewed 2016 node positive breast cancer

patients treated with mastectomy in four randomizedcontrolled chemotherapy trials. They found a local-regional recurrence rate of 21% at 10 years [10]. Local-regional recurrence detected after systemic metastaseswas not routinely recorded, so that the actual number maybe higher. Similarly, in the recent meta-analysis by theEarly Breast Cancer Trialist's Collaborative Group [11],the local-regional recurrence by year ten after mastect-omy with axillary clearance for women with positivenodes was calculated from the data presented to be 24.9%without radiotherapy and 7.7% with radiotherapy. In bothpapers the surgeons involved were again not character-ized and likely to have been diverse.

The most convincing evidence of `̀ better quality''surgery in our study is the median of 18 nodes in MCVHpatients, vs. 7 and 11 nodes in the Danish and Canadiantrials, respectively. This difference, however, could stillbe attributed to more diligence on the part of thepathologists. It should be noted also, that with only oneaxillary recurrence in this series, the bene®t of adjuvantradiotherapy to the axilla following adequate axillarysurgery needs to be questioned. The ECOG study also

Fig. 1. Variation of postmastectomy local-regional recurrence innode positve patients at MCVH over study period.

Fig. 2. Kaplan±Meier survival by stage for patients in study series.

TABLE II. Site and Frequency of Local-Regional Recurrences ofPatients Treated With Mastectomy for Node Positive BreastCancer at MCVH

No. of recurrencesSite (%)

Chest wall 16 (47)Supraclavicular nodes 10 (29)Multiple sites

Chest wall� axilla 1 (3)Chest wall� supraclavicular nodes 5 (15)Supraclavicular� subclavicular nodes 1 (3)Supraclavicular� internal mammary nodes 1 (3)

Total 34 (100)

Fig. 3. Kaplan±Meier local-regional recurrence for all patients andfor stage IIb patients alone in study series.

Node Positive Postmastectomy Recurrence 5

found few axillary recurrences. They reported a medianof 15 axillary nodes. Adjuvant radiotherapy to the axillawas given in both the Danish and Canadian trials. Arecent paper from the Danish group does question theneed for axillary irradiation when adequate surgery isperformed [12]. The majority of local-regional recur-rences in this study as well as in the Danish, Canadian,and ECOG papers were in the chest wall.

Strategies to reduce the effects of local-regional recur-rence include its prevention, through the use of moreeffective adjuvant systemic therapies and/or adjuvantradiotherapy, and the use of salvage treatments, such assurgery and therapeutic radiotherapy. The effects ofsystemic therapies on local-regional recurrence are likelymoderate at best from studies reporting local-regionalrecurrence in node positive mastectomy groups treatedwith and without systemic therapy [13,14]. Studies ofpatients experiencing local-regional recurrences follow-ing modi®ed radical mastectomy have reported salvagerates of up to 86%, largely through the use of surgery and/or radiotherapy [15,16]. In the Danish trial, systemicchemotherapy was used in all patients and salvage wasattempted in patients with local-regional recurrence usingsurgery plus radiotherapy in the control arm and surgeryalone in the radiotherapy arm; salvage rates were notreported. Despite this universal use of systemic therapyand aggressive treatment of local-regional recurrenceswith curative intent, adjuvant radiotherapy not onlyprevented local-regional recurrence, but also provided aclinically and statistically signi®cant 9% survival bene®tat 10 years. The Canadian trial also used systemicchemotherapy in all patients, but did not discuss howlocal-regional recurrences were managed. Similarly,adjuvant radiotherapy showed bene®ts not only forlocal-regional recurrence, but also a survival bene®t of8% at 15 years, this was not statistically signi®cant. Thus,adjuvant systemic therapies are considered insuf®cientin their effects on local-regional recurrence, and although

local-regional recurrences can be salvaged in patients thathave undergone mastectomy, higher rates of local-regional recurrence appear to have a clinically signi®cantdetrimental effect on survival. Local-regional recurrencesin node positive mastectomy patients should be preventedfrom occurring, and the best modality currently availableappears to be adjuvant radiotherapy.

In our univariate and multivariate analyses we wereunable to identify any groups that might have a betteroutcome with respect to local-regional recurrence,however, as indicated previously, the numbers in thisstudy are small.

There is always concern about bias in a retrospectivestudy. These data show no temporal trends and arereasonably representative of survival from the publishedliterature. In a more ideal world, all patients at MCVH,not just those who entered into clinical trials, would havebeen included in this study and the data collectedprospectively. In the absence of this information, thisretrospective review is the best evidence available toprovide for a more balanced approach to determine theapplicability of recommendations for postmastectomyradiotherapy in node positive patients to our own patientpopulation. This analysis cannot be used to support thebene®ts of radiotherapy for either local-regional recur-rence or survival. It can only tell us that the local-regionalrecurrence rate in our patients was not different fromcontrol patients treated similarly in the two publishedtrials.

CONCLUSIONS

There is no evidence in the case series presented here,despite all surgery being performed by fellowship trainedsurgical oncologists and some evidence of `̀ betterquality'' surgery, which would challenge the recommen-dations from the Danish and Canadian trials. Theargument that local-regional recurrence in excess of

TABLE III. Summary of Local-Regional Recurrence Data in Node Positive Women Treated WithMastectomy Without Radiotherapy Discussed in This Paper

Inclusion No. of Local-regional Follow-upStudy criteria patients studied recurrence % (years)

MCV IIb 88 26 9.8 (median)IIb, IIIa 137 27

Canadian (controls) [2] IIb 154 25 10Danish (controls) [1] Node positive (84%) 856 32 10

primary >5 cm (16%)ECOG [10] Node positive 2016 21a 12.1 (median)Early breast cancer Node positive 3339 25a 10

Trialist's CollaborativeGroup [11]

aDid not include local-regional recurrence following diagnosis of distant metastases.

6 Latosinsky and Bear

25% at 10 years following mastectomy in node positivewomen re¯ected inconsistent or variable surgical techni-que in these trials and thus allowed for the bene®ts seenwith radiotherapy is not supported by our data. Womenwith node positive breast cancer treated with mastectomyby surgical oncologists should be considered for adjuvantradiotherapy.

REFERENCES

1. Overgaard M, Hansen PS, Overgaard J, et al.: Post-operativeradiotherapy in high-risk pre-menopausal women with breastcancer who receive adjuvant chemotherapy. N Engl J Med 1997;337:949±955.

2. Ragaz J, Jackson SM, Le N, et al.: Adjuvant radiothe-rapy and chemotherapy in node-positive pre-menopausalwomen with breast cancer. N Engl J Med 1997;337:956±962.

3. Fleming ID, Cooper JS, Henson DE, et al.: (eds): `̀ AJCC CancerStaging Manual.'' Philadelphia: Lippincott-Raven, 1997;171±180.

4. Marks LB, Prosnitz LR: `̀ One to three'' or `̀ four or more''?Selecting patients for postmestectomy radiation therapy. Cancer1997;79:668±670.

5. Hillner BE, Smith TJ, Desch CE: Hospital and physicianvolume or specialization and outcomes in cancer treatment:importance in quality of cancer care. J Clin Oncol 2000;18(11):2327±2340.

6. Hermanek P, Wiebelt H, Staimmer D, et al.: Prognostic factors ofrectum carcinomaÐexperience of the German Multicentre StudySGCRC. German Study Group Colo-Rectal Carcinoma. Tumori1995;81:60±64.

7. Porter GA, Soskolne CL, Yakimets WW, et al.: Surgeon-relatedfactors and outcome in rectal cancer. Ann Surg 1998;227:157±167.

8. Saettler EB, Temple WJ: The surgeon as a prognostic factor in themanagement of pancreatic cancer. Surg Oncol Clin N Am 2000;9(1):133±142.

9. Gillis CR, Hole DJ: Survival outcome of care by specialistsurgeons in breast cancer: a study of 3,786 patients in the west ofScotland. BMJ 1996;312:145±148.

10. Recht A, Gray R, Davidson NE, et al.: Local-regional failure 10years after mastectomy and adjuvant chemotherapy with orwithout tamoxifen without irradiation: experience of the EasternCooperative Oncology Group. J Clin Oncol 1999;17:1689±1700.

11. Early Breast Cancer Trialists' Collaborative Group: Favourableand unfavourable effects on long-term survival of radiotherapy forearly breast cancer: an overview of the randomised trials. Lancet2000;355(9217):1757±1770.

12. Hojris I, Andersen J, Overgaard M, et al.: Late treatment-relatedmorbidity in breast cancer patients randomized to postmastec-tomy radiotherapy and systemic treatment vs. systemic treatmentalone. Acta Oncol 2000;39(3):355±372.

13. Fowble B: Postmastectomy radiation: then and now. Oncology1997;11:213±234.

14. Bonadonna G, Valagussa P, Moliterni A, et al.: Adjuvantcyclophosphamide, methotrexate, and ¯uorouracil in node-positive breast cancer: the results of 20 years of follow-up.N Engl J Med 1995;332:901±906.

15. Willner J, Kiricuta IC, Kolbl O: Local-regional recurrence ofbreast cancer following mastectomy: always a fatal event? Resultsof univariate and multivariate analysis. Int J Radiat Oncol BiolPhys 1997;37:853±863.

16. Chaudary MA, Tong D, Millis R, et al.: Local-regional recurrencefollowing mastectomy for early breast carcinoma: ef®cacy ofradiotherapy at the time of recurrence. Eur J Surg Oncol 1997;23:348±353.

Node Positive Postmastectomy Recurrence 7