PEER-REVIEW REPORTS

Validity of Prognostic Grading Indices for Brain Metastasis Patients Undergoing Repeat Radiosurgery TUMOR

Masaaki Yamamoto1,2, Takuya Kawabe1,3, Yoshinori Higuchi 4, Yasunori Sato5, Tadashi Nariai 6, Shinya Watanabe1,7, Bierta E. Barfod1, Hidetoshi Kasuya 2

Key words Brain metastases - Prognostic index - Radiosurgery - Recurrence

- OBJECTIVES:

Abbreviations and Acronyms BM: Brain metastasis BSBM: Basic Score for Brain Metastases DS-GPA: Diagnosis-Specific Graded Prognostic Assessment GPA: Graded Prognostic Assessment KPS: Karnofsky Performance Status MRI: Magnetic resonance imaging MST: Median survival time RPA: Recursive Partitioning Analysis SIR: Score Index for Radiosurgery SRS: Stereotactic radiosurgery WBRT: Whole-brain radiotherapy

- METHODS:

-

From the 1Katsuta Hospital Mito GammaHouse, Hitachi-naka, Japan; 2Department of Neurosurgery, Tokyo Women’s Medical University Medical Center East, Tokyo, Japan; 3Department of Neurosurgery, Kyoto Prefectural University of Medicine Graduate School of Medical Sciences, Kyoto, Japan; 4Department of Neurological Surgery, Chiba University Graduate School of Medicine, Chiba, Japan; 5Clinical Research Center, Chiba University Graduate School of Medicine, Chiba, Japan; 6 Department of Neurosurgery, Graduate School, Tokyo Medical and Dental University School of Medicine, Tokyo, Japan; and 7Department of Neurosurgery, Faculty of Medicine, University of Tsukuba, Tsukuba, Japan To whom correspondence should be addressed: Masaaki Yamamoto, M.D. [E-mail: [email protected]] Citation: World Neurosurg. (2014) 82, 6:1242-1249. http://dx.doi.org/10.1016/j.wneu.2014.08.008 Journal homepage: www.WORLDNEUROSURGERY.org Available online: www.sciencedirect.com 1878-8750/$ - see front matter ª 2014 Elsevier Inc. All rights reserved.

INTRODUCTION Stereotactic radiosurgery (SRS), with or without whole-brain radiotherapy (WBRT), has emerged as an important modality for managing patients with newly diagnosed brain metastases (BMs). Aoyama et al’s (1) randomized controlled trial comparing treatment results between SRS plus WBRT and SRS alone groups indicated the former to be superior to SRS alone for suppressing

1242

www.SCIENCEDIRECT.com

We tested the validity of 5 prognostic indices, Recursive Partitioning Analysis (RPA), Score Index for Radiosurgery (SIR), Basic Score for Brain Metastases (BSBM), Graded Prognostic Assessment (GPA), and Modified-RPA, for patients who underwent repeat stereotactic radiosurgery (re-SRS).

For this study, we used our database, which included 804 patients who underwent gamma knife re-SRS during the period 1998e2013.

- RESULTS:

There were statistically significant survival differences among patients stratified into 3 or 4 groups based on the 5 systems (P < 0.001). With RPA, SIR, BSBM, and the Modified-RPA, there were statistically significant median survival time (MST) differences between any 2 pairs within the 3/4 groups. With the GPA system, however, the MST difference between the GPA 3.5e4.0 and GPA 3.0 groups did not reach statistical significance (P [ 0.48). There were large patient number discrepancies among the 3/4 groups in the RPA, SIR, BSBM, and GPA whereas patient numbers were very similar among the 3 Modified-RPA system groups. Our present results show the RPA and BSBM systems to reflect changes less well, with 86%e 95% of patients remaining in the same categories between the first and second SRS procedures. However, with SIR, GPA, and the Modified-RPA, 25%e31% of patients were categorized into different subclasses, either better or worse. With the modified-RPA system, such categorical change correlated well with post-re-SRS MSTs.

- CONCLUSIONS:

Among the 5 systems, based on patient number proportions, MSTseparation among the 3/4 groups, and/or detailed reflection of status changes, the Modified-RPA system was shown to be most applicable to re-SRS patients.

newly appearing tumors, although there was no significant median survival time (MST) difference between the 2 groups. Debate persists, however, as to whether WBRT is necessary for all patients with BMs. The primary argument against WBRT stems from the risk of deterioration of neurocognitive function, which cannot be ignored in long-surviving patients (2). Second, because WBRT generally is considered to be unrepeatable, the availability of an alternative treatment for BMs allows WBRT to be reserved for subsequent treatment attempts, i.e., in cases with meningeal dissemination or miliary metastases for which only WBRT is effective. Therefore, an increasing number of patients with BM have been treated with SRS alone. In patients with BM who undergo SRS alone, new BMs inevitably appear with

relatively high incidences during post-SRS follow-up. Hanssens et al (7) recently reported that, using high-performance magnetic resonance imaging (MRI), new tumors were diagnosed in 40% of 835 patients with BM who had undergone SRS alone. Recently published studies based on more than 1000 patients with BM treated with SRS alone disclosed that reSRS for new tumors was required in 22%e 34% of all cases (9, 17, 18, 34). Several retrospective studies, based on relatively small patient numbers, have documented re-SRS to be safe and effective (3, 13, 19, 28). However, although the 5 prognostic grading indices discussed herein for initially treated patients with BM are well established (5, 15, 20, 27, 31, 34, 36), none was developed for or has yet been validated in the re-SRS setting.

WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2014.08.008

PEER-REVIEW REPORTS MASAAKI YAMAMOTO ET AL.

MATERIAL AND METHODS Study Population This was an institutional review boarde approved, retrospective cohort study using our prospectively accumulated database including 2825 consecutive patients who underwent SRS alone using a Gamma Knife, without WBRT, for BMs at the Katsuta Hospital Mito GammaHouse during the 15-yearperiod between July 1998 and June 2013 (Tokyo Women’s Medical University; institutional review board #1981). Among the 2825 patients, excluding 37 patients who underwent WBRT before SRS, we studied 804 (28.5%) undergoing re-SRS mostly for newly developed lesions only (682 patients, 84.8%) and, uncommonly, recurrence of treated lesions (122, 15.2%). Table 2 summarizes clinical characteristics before re-SRS. All patients had been referred to us for SRS by their primary physicians. Therefore, patient selections had mostly been made outside of our facilities. Patient selection criteria may well have differed somewhat among the referring physicians. Therefore, the first author (M.Y.) ultimately decided whether a patient would be accepted for SRS in all cases. We did not perform SRS on patients with low Karnofsky Performance Status (KPS) scores (10) attributable to systemic diseases, patients who were noncooperative due to poor neurocognitive function, who had meningeal dissemination, and/or who had an expected survival period of 3 months or less. Therefore, only 4.2% of the 804 patients were categorized into RPA class 3 (5). Also, the treating physicians responsible for each patient decided the indications for both surgery and radiotherapy. The treatment strategy was explained in detail to each patient and at least one of their adult relatives by the first author

Table 1. Outline of Reported Grading Indices for Patients with Brain Metastases (BMs)

Table 1. Continued Basic Score for Brain Metastases (BSBM) (15)

Recursive Partitioning Analysis (RPA) (5) Class 1

BSBM Scoring Criteria

Age 65 vs. 65 years) and post-re-SRS survival. Table 5 shows post-re-SRS MSTs according to tumor number groups, i.e., 1e4 and
5 for the 5 grading indices. In patients with 1e4 tumors, a better score is associated with a longer post-re-SRS MST, and the survival difference by 3- or 4-group stratification is statistically significant for all 5 indices (P < 0.001). However, in patients with
5 tumors, the same results were obtained for the 4 indices, i.e., the RPA, SIR, BSBM, and modified-RPA systems (P < 0.001), but not for the GPA system (P ¼ 0.13). DISCUSSION After publication of the index based on RPA by Gaspar et al. (5), 4 other indices, i.e., SIR, BSBM, GPA, and the ModifiedRPA, were proposed for patients with newly diagnosed BMs (15, 20, 27, 31, 34, 36). However, to our knowledge, there have been no reports discussing a prognostic grading index for patients with newly appearing or recurrent BMs after the initial SRS. We applied these 5 indices to

WORLD NEUROSURGERY 82 [6]: 1242-1249, DECEMBER 2014

our patients undergoing re-SRS. As shown in Figure 1, 4 (RPA, SIR, BSBM, and the Modified-RPA) of the 5 indices showed statistically significant post-re-SRS MST differences among the 3 or 4 subgroups without, or with only minimal, overlapping of 95% CIs between any 2 pairs of groups (P < 0.05 for all subclasses). With the GPA system, however, the post-re-SRS MST difference between the GPA 3.5e4.0 and GPA 3.0 groups did not reach statistical significance (P ¼ 0.48). However, it should be noted that there were some biases in patient selection in our patient series. As stated in the Materials and Methods section, we did not perform SRS on patients with low KPS scores (10) due to systemic diseases, which may have influenced our results. Although post-re-SRS MSTs differed significantly among 3 or 4 subclasses based on the 3 indices (RPA, SIR, and BSBM), there were large discrepancies in patient numbers among these subclasses. With the RPA, 83.7% of all patients were class 2, whereas 4.2% were class 3, the dissociation rate being 19.9 versus 1.0. In the same way, dissociation rates between 2 groups, those with the greatest and lowest numbers of patients, were 7.0 versus 1.0

www.WORLDNEUROSURGERY.org

1245

PEER-REVIEW REPORTS MASAAKI YAMAMOTO ET AL.

GRADING MET PATIENTS UNDERGOING RE-SRS

TUMOR Figure 2. Category changes between the first and the second treatment according to the 5 grading indices. RPA, Recursive Partitioning Analysis (5); SIR, Score Index for Radiosurgery (27); BSBM, Basic Score for Brain

with the SIR and 15.2 versus 1.0 with the BSBM. These large number discrepancies in patients between classes or groups might reflect clinical factors and survival periods varying markedly within subsets including those with the largest patient numbers. A prognostic index with such large patient number discrepancies among subgroups would, in our view, be fundamentally inadequate. Furthermore, neither the RPA nor the BSBM incorporates BMrelated factors, i.e., tumor number or tumor size, both of which are widely accepted as major factors influencing patient survival periods. The SIR requires the volume of the largest lesion. Volume estimation is commonly performed at the time of SRS or stereotactic radiotherapy, but not always at the time of WBRT. Therefore, this index is rarely used clinically because treatment options must be

1246

www.SCIENCEDIRECT.com

Metastases (15); GPA, Graded Prognostic Assessment (20); and Modified-RPA (34, 36).

selected for most patients without knowing actual tumor volumes. As discussed previously, all 4 previously proposed systems, i.e., the RPA, SIR, BSBM, and GPA, have limitations when applied to our set of patients undergoing re-SRS. With the Modified-RPA system, however, there were statistically significant MST differences among the 3 groups without overlapping of 95% CIs between any 2 pairs of groups. Additionally, with this system, patient number discrepancies among the 3 groups were minimal, such that this system can be regarded as an adequate prognostic index for patients requiring salvage treatment. Because a patient’s general status and BM condition usually change during the observation period, another important issue for prognostic grading indices is that these changes be reflected accurately. The present

study showed the RPA and BSBM systems to reflect changes rather poorly, with 95.2% and 85.8%, respectively, of patients remaining in the same categories between the 2 procedures (Table 3). However, with the other 3 systems, 25%e31% of patients were categorized into different subclasses, either better or worse. Thus, these three systems are considered to be useful for meticulously evaluating changes in a patient’s general condition as well as BM status. As shown in Table 3, such categorical changes correlated well with post-re-SRS MSTs when using the 4 indices, i.e., the RPA, SIR, BSBM, and Modified RPA systems. Regardless of which of the 4 indices are used, patients in worse categories are not good candidates for SRS because of their anticipated relatively short survival periods. Recently, Sperduto et al. (21) modified their original GPA system and developed a

WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2014.08.008

PEER-REVIEW REPORTS MASAAKI YAMAMOTO ET AL.

GRADING MET PATIENTS UNDERGOING RE-SRS

Table 3. Summary of Category Changes Between the First and Second SRS and Their Impact on Survivals After the Second Procedure for Each of the Five Grading Indices

Indices

MST (95% CI)

Better (B)

Unchanged (U)

Worse (W)

B and U

W

HR (95% CI)

P Value

6 (0.7%)

765 (95.2%)

33 (4.1%)

7.3 (6.8e8.9)

2.7 (1.2e5.1)

0.607 (0.435e0.877)

< 0.009

SIR

79 (9.8%)

606 (75.4%)

119 (14.8%)

7.3 (6.9e8.2)

5.5 (4.4e6.7)

0.701 (0.576e0.861)

< 0.001

BSBM

42 (5.2%)

690 (85.8%)

72 (9.0%)

7.4 (6.9e8.2)

4.2 (2.7e5.3)

0.516 (0.408e0.661)

< 0.001

RPA class

GPA

92 (11.4%)

550 (68.5%)

162 (20.1%)

7.3 (6.6e8.0)

6.5 (5.5e8.4)

0.904 (0.759e1.085)

0.27

Modified-RPA class

82 (10.2%)

572 (71.1%)

150 (18.7%)

7.3 (6.6e8.1)

6.5 (5.3e7.6)

0.810 (0.677e0.976)

0.03

SRS, stereotactic radiosurgery; MST, median survival time; CI, confidence interval; RPA, Recursive Partitioning Analysis (5); SIR, Score Index for Radiosurgery (26); BSBM, Basic Score for Brain Metastases (15); GPA, Graded Prognostic Assessment (20).

new index, the DS-GPA; a user-friendly worksheet of this system was demonstrated (23). Their new system uses different

scoring that takes into account different primary tumor types, as described herein, because it is widely known that there are

Table 4. Uni- and Multivariable Analyses of Survival After Repeat Stereotactic Radiosurgery Univariable Analyses Factors

Multivariable Analyses

HR (95% CI)

P Value

HR (95% CI)

P Value

1.296 (1.119e1.503)

< 0.001

1.387 (1.194e1.614)

< 0.001

Continuous

1.007 (1.000e1.014)

0.046

>65 vs. 65 years

1.097 (0.947e1.269)

0.22

Sex Male vs. female Age

1.095 (0.943e1.270)

0.24

Tumor numbers Continuous

1.034 (1.025e1.042)

< 0.001

Solitary vs. multiple

0.549 (0.456e0.658)

< 0.001

0.666 (0.539e0.820)

< 0.001

4 vs
5

0.566 (0.488e0.658)

< 0.001

0.621 (0.524e0.734)

< 0.001

0.585 (0.504e0.680)

< 0.001

0.637 (0.547e0.741)

< 0.001

1.323 (1.141e1.533)

< 0.001

1.371 (1.179e1.593)

< 0.001

0.460 (0.379e0.563)

< 0.001

0.457 (0.373e0.564)

< 0.001

Continuous

1.016 (1.008e1.023)

< 0.001