Clinical Oncology (2008) 20: 337e339doi:10.1016/j.clon.2008.03.005

Original Article

Prediction of Very Short Survival in Patients with BrainMetastases from Breast Cancer

C. Nieder*y, K. Marienhagenz, R. Thammx, S. T. Astnerx, M. Mollsx, J. Norumyz*Radiation Oncology Unit, Nordlandssykehuset HF, 8092 Bodø, Norway; yInstitute of Clinical Medicine, Faculty of Medicine,University of Tromsø, Tromsø, Norway; zDepartment of Oncology, University Hospital of North Norway, Tromsø, Norway;xDepartment of Radiation Oncology, Klinikum rechts der Isar der Technischen Universitat Munchen, 81675 Munich, Germany

ABSTRACT:Aims: Current prognostic models are not accurate enough to identify brain metastases patients with very short survival,i.e. !2 months, who are unlikely to derive major benefit from whole brain radiotherapy. Our aim was to develop a morereliable model.Materials and methods: This was a retrospective analysis of a German database, which was used to develop a score, andan additional database from Norway, which was used for validation purposes.Results: The groups included 67 and 32 patients, respectively. An analysis of prognostic factors resulted in a risk scorebased on performance status, extra-cranial metastases, the interval from breast cancer to brain metastases and a needfor corticosteroid treatment, which classified 63 of 67 test patients correctly. However, the validation failed andunfortunately the risk score that performed best in the Norwegian patients (31 of 32 correctly predicted) was notapplicable to the German patients.Conclusions: The prediction of short survival is associated with several caveats and seems to result in an unacceptablerisk of withholding radiotherapy in patients who actually survive for longer than 2 months. Nieder, C. et al. (2008).Clinical Oncology 20, 337—339

ª 2008 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.

Key words: Brain metastases, breast cancer, prognosis, radiotherapy, whole brain radiotherapy

Introduction

Whole brain radiotherapy (WBRT) continues to represent animportant palliative treatment option for patients withbrain metastases from breast cancer. The fact that a certainpercentage of these patients have very short survival timesafter WBRT suggests that accurate survival predictionmodels might help to avoid overtreatment [1]. Even inbrain metastases trials with restrictive inclusion criteria,more than 10% of the patients died within 8 weeks [2,3]. Ithas been reported that survival after corticosteroids aloneis in the order of 6e8 weeks (overview in [4]). In groupswith such short survival time, adding WBRT is unlikely tobenefit the patients to a meaningful extent. Prognosticscores such as the recursive partitioning analysis (RPA)classes can support decision making and treatment recom-mendations in patients with brain metastases [5,6].However, survival in the most unfavourable class III is quitevariable, with 40e50% of patients dying within 2 months,but 15e20% surviving for more than 6 months [5,6].Therefore, we aimed at developing an improved modelthat predicts the survival time !2 months in patients withbrain metastases from breast cancer and is based on readilyavailable clinical parameters.

0936-6555/08/200337þ03 $35.00/0 ª 2008 The Royal Col

Materials and Methods

A German database of 67 patients with brain metastasesfrom breast cancer treated with WBRT (10 fractions of 3 Gy)was used for the development of a prognostic score. Nosurgery or radiosurgery was used in this group, except forsalvage treatment after WBRT failure. Potential prognosticfactors were examined (RPA class, Karnofsky performancestatus [KPS], age, presence of extra-cranial metastases,interval from primary diagnosis to brain metastases, needfor corticosteroid treatment to control neurological symp-toms, presence of a single brain metastasis, etc.). Thepresence of an uncontrolled primary tumour was notevaluated separately, as this was very uncommon and isunlikely to cause death, in contrast to, e.g. uncontrolledadvanced lung cancer. Information on primary tumourfeatures, such as grading, hormone receptor status andHER2 receptor status, was not taken into consideration as itwas not available in both the test and the validation groups.Lymphopenia was not available either. It was first examinedwhether RPA class alone would predict survival !2 months.Then a score predicting more reliably such a short survivalwas developed based on the significant prognostic factors.A group of identically treated patients from Tromsø,

lege of Radiologists. Published by Elsevier Ltd. All rights reserved.

338 CLINICAL ONCOLOGY

Norway, was examined to validate the score. The KaplaneMeier method was used to generate actuarial survivalcurves. These were compared with the Log-rank test toobtain univariate prognostic factors (all carried out usingthe SPSS software). A P value !0.05 was consideredstatistically significant. Multivariate analysis of prognosticfactors was not carried out, as the groups were small.

Results

The test group consisted of 67 patients. Table 1 shows thedemographics. The significant prognostic factors for sur-vival were RPA class, KPS, presence of extra-cranialmetastases, interval from the diagnosis of breast cancerto brain metastases, and minimum dose of dexamethasoneneeded to control the patient’s symptoms during WBRT.The three RPA classes had median survival times of 19, 8and 1.5 months, respectively. However, survival !2 monthswas not well predicted by the presence of RPA class III.On the one hand, 17/21 patients who died within the first2 months belonged to RPA class III and the other ones toclass II. On the other hand, 13/46 patients with survival O4months also belonged to RPA class III. Therefore, a newprognostic score was developed. It was based on thesignificant prognostic factors KPS, extra-cranial metasta-ses, interval and dose of dexamethasone treatment. Table2 shows the prognostic score that fitted this data set best.This score predicted the outcome of 63/67 patientscorrectly when the cut-off for survival !2 months was setto 6 points. Two incorrect cases had a survival time of lessthan 2 months, but only 5 points, i.e. were predicted tosurvive longer. WBRT might thus represent overtreatmentin these two patients. The other two incorrect patientswere assigned R 6 points, i.e. predicted to die within 2months, but survival was 3e5 months.

The 32 patients from the Tromsø database were thenused to validate the score. Four of these patients hadsurvival !2 months and three were predicted correctly.Twenty-eight patients had longer survival. Of these, only 23

Table 1 e Patient characteristics

German group (n

Median age (years), range 57, 33e71Median KPS (%), range 70, 40e90Median time interval (months), range* 38, 6e120% Single brain metastasis 31% Without extra-cranial metastases 19% Without corticosteroids 18% With O20 mg dexamethasone/day 27% RPA class I vs II vs III 18 : 37 : 45% Chemotherapy before brain metastases diagnosis 70% Haemoglobin ! 11.5 g/dl 21% Albumin ! 32.5 g/l 13% Haemoglobin ! 11.5 g/dl and albumin ! 32.5 g/l 9

KPS, Karnofsky performance status; RPA, recursive partitioning analysis.

were predicted correctly. In the other five patients, 6e7points were assigned, but survival was up to 14 months.There was thus a high risk for withholding WBRT based onthe predicted prognosis. With only four patients survivingfor !2 months, the Tromsø database was not very useful fordeveloping an alternative risk score. Nevertheless, a modelwas developed, which ultimately predicted survival cor-rectly in 31 of these 32 patients (Table 3). It consisted offour factors (KPS !70, extra-cranial metastases, interval!30 months and the combination of low haemoglobin pluslow albumin), resulting in a maximum of 4 possible riskpoints. Patients with 3e4 points were predicted to survivefor !2 months. One patient who survived for 14 months wasassigned 3 points and thus incorrectly predicted. Whenapplying this score to the German database, the number ofcorrectly predicted patients dropped from 63 with the firstmodel to only 56. As also observed in the German data set,RPA class was unable to predict short survival. All fourpatients surviving for !2 months belonged to class III.However, seven other patients also belonged to class III.Their maximum survival was 12 months.

Discussion

Outside prospective clinical trials that include selectedpatients only, many of those with brain metastases frombreast cancer die within 2 months, e.g. 20e30% in the seriesreported by Mahmoud-Ahmed et al. [7], Bartsch et al. [8]and Le Scodan et al. [9]. We attempted to develop andvalidate a clinical risk score that might help to avoidovertreatment in such patients, if one assumes that patientsdying within 8 weeks can be managed adequately by bestsupportive care focused on corticosteroid treatment.However, with the available patient groups and potentialprognostic factors, validation of the candidate scores wasnot successful. Whether the use of larger databases or theinclusion of other parameters might improve our ability topredict very short survival remains to be shown. Therecently suggested prognostic factor lymphopenia [9,10]

¼ 67) Norwegian group (n¼ 32) Significant prognostic factor

55, 34e81 No70, 50e90 Yes52, 3e216 Yes

16 No9 Yes

12.5 Only in German group22 Only in German group

6 : 59 : 34 Yes81 No28 No9 No6 Only in Norwegian group

*Interval from breast cancer diagnosis to brain metastases.

Table 2 e Final score predicting survival !2 months and mediansurvival for the different groups in the German database

Parameter Points

Karnofsky performancestatus

0 if O70, 1 if 70 or 60,2 if 50 or less

Extra-cranial metastases 1 if present*, 0 if not presentCorticosteroid treatment 0 if no steroids were given,

1 if daily dexamethasone dosewas 1e20 mg, 2 if O20 mg

Time interval betweenbreast cancer and brainmetastases

0 if O2 years, 1 if between 1 and2 years, 2 if 1 year or shorter

Risk score points Median survival in months

0e1 (n¼ 5) 49.02 (n¼ 6) 32.33 (n¼ 10) 10.14 (n¼ 11) 5.25 (n¼ 14) 3.66e7 (n¼ 21) 1.2

*0 if patients had only skin metastases; 0 if patients had only lymphnode metastases staged as distant rather than nodal metastases,e.g. neck or contralateral axilla.

339BRAIN METASTASES FROM BREAST CANCER

has not been routinely assessed in the patients analysedhere and could therefore not be included. Lock et al. [1]previously evaluated the same end point in a larger group ofpatients, but with different primary tumours. They foundthe Eastern Cooperative Oncology Group performancestatus and the number of metastatic sites to be importantparameters. However, their model classified 68% of patientscorrectly, but 55% would have been incorrectly predicted todie early. Our group has also attempted to predict survival

Table 3 e Final score predicting survival !2 months and mediansurvival for the different groups in the Norwegian database

Parameter Points

Karnofsky performance status 0 if O60, 1 if 60 or lessExtra-cranial metastases 1 if present*, 0 if not presentHaemoglobin ! 11.5 g/dl andalbumin ! 32.5 g/l

1 if present, 0 if not present

Time interval between breastcancer and brain metastases

0 if R 2.5 years,1 if !2.5 years

Risk score points Median survival in months

0e1 (n¼ 17) 9.12 (n¼ 11) 4.03e4 (n¼ 4) 1.5

*0 if patients had only skin metastases; 0 if patients had only lymphnode metastases staged as distant rather than nodal metastases,e.g. neck or contralateral axilla.

!2 months in patients with brain metastases from non-small cell lung cancer, but arrived at the same conclusion. Ifit is not possible to avoid overtreatment in some patientswith very unfavourable survival, short-course WBRT shouldbe considered in patients presenting with several adverseprognostic factors [11].

Author for correspondence: C. Nieder, Radiation Oncology Unit,Medical Department, Nordlandssykehuset HF, Prinsensgate 164,8092 Bodø, Norway. Tel: þ47 7557 8449; E-mail: carsten.nieder@nlsh.no

Received 10 July 2007; received in revised form 5 February 2008;accepted 10 March 2008

References

1 Lock M, Chow E, Pond GR, et al. Prognostic factors in brainmetastases: can we determine patients who do not benefit fromwhole-brain radiotherapy? Clin Oncol (R Coll Radiol) 2004;16:332e338.

2 Andrews DW, Scott CB, Sperduto PW, et al. Whole brain radiationtherapy with and without stereotactic radiosurgery boost forpatients with one to three brain metastases: phase III results ofthe RTOG 9508 randomized trial. Lancet 2004;363:1665e1672.

3 Mehta MP, Rodrigus P, Terhaard CH, et al. Survival andneurologic outcomes in a randomized trial of motexafingadolinium and whole-brain radiation therapy in brain meta-stases. J Clin Oncol 2003;21:2529e2536.

4 Khuntia D, Brown P, Li J, Mehta MP. Whole-brain radiotherapy inthe management of brain metastasis. J Clin Oncol 2006;24:1295e1304.

5 Gaspar L, Scott C, Rotman M, et al. Recursive partitioninganalysis (RPA) of prognostic factors in three Radiation TherapyOncology Group (RTOG) brain metastases trials. Int J RadiatOncol Biol Phys 1997;37:745e751.

6 Nieder C, Nestle U, Motaref B, Walter K, Niewald M, Schnabel K.Prognostic factors in brain metastases: should patients be selectedforaggressive treatmentaccording torecursivepartitioninganalysis(RPA) classes? Int J Radiat Oncol Biol Phys 2000;46:297e302.

7 Mahmoud-Ahmed AS, Suh JH, Lee SY, Crownover RL,Barnett GH. Results of whole brain radiotherapy in patientswith brain metastases from breast cancer: a retrospectivestudy. Int J Radiat Oncol Biol Phys 2002;54:810e817.

8 Bartsch R, Fromm S, Rudas M, et al. Intensified local treatmentand systemic therapy significantly increase survival in patientswith brain metastases from advanced breast cancer e a retro-spective analysis. Radiother Oncol 2006;80:313e317.

9 Le Scodan R, Massard C, Mouret-Fourme E, et al. Brainmetastases from breast carcinoma: validation of the RadiationTherapy Oncology Group recursive partitioning analysis classi-fication and proposition of a new prognostic score. Int J RadiatOncol Biol Phys 2007;69:839e845.

10 Claude L, Perol D, Ray-Coquard I, et al. Lymphopenia: a newindependent prognostic factor for survival in patients treatedwith whole brain radiotherapy for brain metastases from breastcarcinoma. Radiother Oncol 2005;76:334e339.

11 Priestman TJ, Dunn J, Brada M, et al. Final results of the RoyalCollege of Radiologists’ trial comparing two different radio-therapy schedules in the treatment of cerebral metastases. ClinOncol (R Coll Radiol) 1996;8:308e315.