Abstract
Purpose:
To update the Diagnosis-Specific Graded Prognostic Assessment (DS-GPA) for a markedly heterogeneous patient population, patients with melanoma and brain metastases, using a larger, more current cohort, including molecular markers.
Methods:
The original Melanoma-GPA is based on data from 483 patients whose conditions were diagnosed between 1985 and 2005. This is a multi-institutional retrospective database analysis of 823 melanoma patients with newly diagnosed brain metastases from January 1, 2006, to December 31, 2015. Multivariable analyses identified significant prognostic factors, which were weighted and included in the updated index (Melanoma-molG-PA). Multiple Cox regression was used to select and weight prognostic factors in proportion to their hazard ratios to design the updated Melanoma-molGPA in which scores of 4.0 and 0.0 are associated with the best and worst prognoses, as with all of the diagnosis-specific GPA indices. Log-rank tests were used to compare adjacent classes.
Results:
There were 5 significant prognostic factors for survival (age, Karnofsky performance status [KPS], extracranial metastases [ECM], number of brain metastases, and BRAF status), whereas only KPS and the number of brain metastases were significant in the original Melanoma-GPA. Median survival improved from 6.7 to 9.8 months between the 2 treatment eras, and the median survival times for patients with Melanoma-molGPA of 0 to 1.0, 1.5 to 2.0, 2.5 to 3.0, and 3.5 to 4.0 were 4.9, 8.3, 15.8, and 34.1 months (P<.000l between each adjacent group).
Conclusions:
Survival and our ability to estimate survival in melanoma patients with brain metastases has improved significantly. The updated Melanoma-molGPA, a user-friendly tool to estimate survival, will facilitate clinical decision making regarding whether and which treatment is appropriate and will also be useful for stratification of future clinical trials. To further simplify use, a free online/smart phone app is available at brainmetgpa.com.
Summary
The Graded Prognostic Assessment (GPA) is a diagnosis-specific prognostic index for patients with brain metastases. The Melanoma-GPA has been updated based on larger and more current data. Additional factors, including BRAF status, have been found to be prognostic. Those factors, weighted by significance, have been incorporated into the new Melanoma-molGPA. Survival and the ability to predict survival for this population has improved significantly, and the Melanoma-molGPA will facilitate clinical decision-making and stratification of clinical trials.
Introduction
Brain metastases are a common and complex conundrum for cancer care. An estimated 300,000 patients receive diagnoses each year of brain metastases in the United States (1), and that incidence is growing because of advances in treatment that result in patients living longer and thus at risk for brain metastases (2). It is a complex problem because of the marked heterogeneity of this patient population and the wide range of prior treatments (none vs extensive) they may have received at the time of diagnosis of the brain metastases. Of all malignancies, melanoma has not only the highest propensity to spread to the brain (3) but also rapidly expanding treatment options (surgery, stereotactic radiosurgery [SRS], whole brain radiation therapy [WBRT], targeted drug therapies, and immunotherapies [4]) resulting in greater heterogeneity. This heterogeneity has long plagued interpretation of clinical trials involving this patient population because it was essentially impossible to sufficiently stratify studies to verify that similar groups of patients were being compared.
This problem led to efforts to better understand prognosis, and progress was made. Gaspar et al (5) published the Radiation Therapy Oncology Group (RTOG) Recursive Partitioning Analysis for brain metastases in 1997. This prognostic index consisted of 3 classes: class I (age <65, Karnovsky performance status [KPS] ≥70, controlled primary tumor, no extracranial metastases [ECM]), class II (all patients not in classes I or III), and class III (KPS <70), which correlated with median survival times of 7.7, 4.5 and 2.3 months, respectively.
Our group published a new more quantitative prognostic index, the Graded Prognostic Assessment (GPA) in 2008 based on age, KPS, ECM, and number of brain metastases (6). Those prognostic factors were weighted by regression coefficients and scaled in such manner that patients with the best/worst prognosis would have a GPA of 4.0/0.0, respectively. In 2010, we refined the GPA to be diagnosis specific when we found that survival varies by diagnosis and diagnosis-specific prognostic factors (7). The Breast-GPA was then further refined using tumor subtype (8) and a summary report was published (9). More recently, the Lung-GPA was updated using molecular factors (EGFR/ALK status) (10, 11), respectively.
The original Melanoma-GPA found that only 2 factors were significant (KPS and the number of brain metastases). We recently published the effect of gene mutations on survival in melanoma patients with brain metastases (12). The purpose of this study was to update the Melanoma-GPA using molecular markers and a larger sample size from the current treatment era.
Methods
A multi-institutional retrospective institutional review board—approved database was created, comprising 823 melanoma patients with newly diagnosed brain metastases from 2006 to 2015. Variables considered included the 4 in the existing GPA: KPS, age, presence of extracranial metastases, and number of brain metastases. Additional variables included gene mutation status (BRAF, CKIT, and NRAS ), sex, volume of brain metastases (for those treated with SRS), and time from primary diagnosis to brain metastases. Race was not included because of the low percentage of non white patients. Type of treatment was not considered because the purpose of a prognostic index is to estimate survival before treatment.
Statistical analyses
Multiple Cox regression was used to initially select and weight variables to be included in the new Melanoma-molGPA. The primary endpoint was overall survival measured from the start of treatment for brain metastases. Continuous variables were categorized to assess potential nonlinear effects. Both effect magnitude (hazard ratio) and statistical significance were used to select variables. Variables with larger hazard ratios were given larger point values. The final index was chosen on the basis of separation of prognostic classes with respect to overall survival, distribution of patients, and simplicity. Log-rank tests were used to compare adjacent classes.
Results
The patient characteristics, survival by gene status (BRAF, CKIT, and NRAS ), risk of death by treatment and treatment era, and a summary of drug therapy for this patient cohort have been previously published (13). The treatment breakdown for the 823 patients was as follows: 56% SRS alone, 12% surgery + SRS, 11% WBRT, 9% WBRT + SRS, 4% surgery + WBRT, and 1% surgery + WBRT + SRS.
Table 1 shows the multivariable model used to select and weight factors in the Melanoma-molGPA. Five prognostic factors were found to be significant for survival in the current study: age, KPS, extracranial metastases, number of brain metastases, and BRAF status, whereas only 2 (KPS and number of brain metastases) were significant in the original Melanoma-GPA study (10). The overall median survival times for patients in this cohort and the original Melanoma GPA study (1985–2005, n = 481) were 9.8 and 6.7 months (P<.001), respectively. Similarly, survival by GPA group also improved: Melanoma-molGPA of 0 to 1.0,1.5 to 2.0,2.5 to 3.0, and 3.5 to 4.0 were 4.9, 8.3, 15.8, and 34.1 months, respectively (P<.0001 between each adjacent group) compared with the prior era (GPA 0–1.0, 1.5–2.0, 2.5–3.0, and 3.5–4.0 were 3.4, 4.7, 8.8, and 13.2 months, respectively). Figure 1 shows a comparison of the Kaplan-Meier curves for the original and the new Melanoma-GPA for survival by GPA class. Both demonstrate clear separation between adjacent classes and together reflect the improvement in survival between the 2 study periods. Notably, the GPA for the worst prognostic group (GPA 0.0–1.0) has increased slightly from 3.4 to 4.9 months whereas the GPA for the best prognostic group (GPA 3.5–4.0) has improved markedly from 13.2 to 34.1 months. The distribution of patients by class shows that 17% of patients are in the worst prognostic group in both the prior and current studies, whereas 23% and 5% of patients were in the best prognostic group in the prior and current studies, respectively.
Table 1.
Hazard ratios of significant prognostic factors used to derive melanoma-GPA
| Variable | n | Hazard ratio | 95% Confidence interval |
|---|---|---|---|
| Age, y | |||
| <70 | 651 | 1.00 | |
| 70–97 | 172 | 1.41 | 1.16–1.71 |
| KPS | |||
| 60–70 | 178 | 2.24 | 1.77–2.83 |
| 80 | 192 | 1.34 | 1.09–1.64 |
| 90–100 | 453 | 1.00 | |
| ECM | |||
| Absent | 136 | 1.00 | |
| Present | 687 | 2.06 | 1.61–2.65 |
| No. of BM | |||
| 1 | 328 | 1.00 | |
| 2–4 | 303 | 1.27 | 1.05–1.55 |
| >4 | 192 | 1.72 | 1.38–2.14 |
| BRAF status | |||
| Positive | 297 | 1.00 | |
| Negative | 287 | 1.30 | 1.06–1.58 |
| Unknown | 239 | 1.94 | 1.52–2.48 |
Abbreviations: BM = brain metastases; ECM = extracranial metastases; GPA = Graded Prognostic Assessment; KPS = Karnofsky performance status.
A multiple Cox regression model for overall survival from start of BM treatment is shown, with hazard ratios of 1.0 representing the reference group of each variable.
Fig. 1.
Kaplan-Meier curves comparing the original and new melanoma-GPA. Abbreviations: BM = brain metastases; GPA = Graded Prognostic Assessment; MS = median survival in months.
The factors analyzed and found not to be significant for survival were gender, volume of brain metastases, CKIT and NRAS status, and time from primary diagnosis to the diagnosis of brain metastases.
Table 2 shows the definition of the updated Melanoma-molGPA in a worksheet format designed to facilitate quick and easy calculation of the GPA score. To further simplify this process, a free online smartphone app has been updated and is available at brainmetgpa.com.
Table 2.
Melanoma GPA worksheet
| Prognostic factor |
GPA scoring criteria |
Patient Score |
||
|---|---|---|---|---|
| 0 | 0.5 | 1.0 | ||
| Age, y | ≥70 | <0 | - | |
| KPS | ≤70 | 80 | 90–100 | - |
| ECM | Present | Absent | - | |
| No. of BM | >4 | 2–4 | 1 | - |
| BRAF gene | Negative/unknown | Positive | - | |
| status | ||||
| Sum | - | |||
Abbreviations: BM = brain metastases; ECM = extracranial metastases; GPA = Graded Prognostic Assessment; KPS = Karnofsky performance status; MS = median survival in months.
MS by GPA: 0–1.0 = 4.9, 1.5–2.0 = 8.3, 2.5–3.0 = 15.8, 3.5–4.0 = 34.1.
Discussion
A better estimate of survival will facilitate the doctor/patient/family discussion of whether and which treatment is appropriate. A patient with a Melanoma-molGPA of 3.5 to 4.0 can expect to live nearly 3 years, and more aggressive treatment may be indicated. On the other hand, a patient who has a GPA of 0.0 to 1.0 has an expected survival of <5 months and therefore may choose hospice. These decisions are difficult, and the cost of hope can be enormous (13). The QUARTZ trial, a randomized trial of dexamethasone and supportive care versus WBRT in lung cancer patients with brain metastases, showed no difference in survival or quality of life (14). Even though WBRT is now used less than in the past, a trend also reflected in the treatment patterns reported here between the 2 eras (1985–2005 and 2006–2015), the QUARTZ trial raises the question of when to treat and when to refer to hospice. This and the other updated diagnosis-specific GPA indices inform that decision (9, 11).
Whereas BRAF-targeted therapy and immunotherapy have clearly had an impact on outcomes in melanoma in general, it remains unresolved whether these agents benefit patients with brain metastases. In fact, our group recently published a review of the relevant literature and the possible explanations for the findings that the extracranial response to these agents exceeds the intracranial response (12).
Limitations
The study is retrospective, so despite the large sample size, there is the inherent selection bias present in all retrospective studies. Accordingly, we cannot conclude that 1 treatment is better than another. Furthermore, the type, timing, combination, and sequence of chemotherapies, targeted therapies, and immunotherapies, both before and after the diagnosis of brain metastases, varied widely, thus precluding any conclusion regarding the efficacy of those agents. Another limitation is the percentage of patients with unknown mutation status (29% BRAF, 65% NRAS, 64% CKIT ), which is mostly due to the timing within our study period that these tests were initiated at the various centers.
In conclusion, this updated prognostic index, the Melanoma-Graded Prognostic Assessment (Melanoma-molGPA) incorporates molecular markers and provides a user-friendly tool to estimate survival for melanoma patients with brain metastases and is useful in clinical decision making and stratification of future clinical trials. Accurate stratification is crucial to interpretation of the results of clinical trials in this markedly heterogeneous patient population.
Acknowledgment—
The authors thank Susan Lowry, Database Programmer/Analyst and REDCap Administrator, Biostatistical Design and Analysis Center, Clinical and Translational Science Institute, University of Minnesota.
Supported by National Institutes of Health (NIH) grant No. UL1TR(X)0114 from the National Center for Advancing Translational Sciences (NCATS), and by NIH grant No. P30 CA77598 using the Biostatistics and Bioinformatics Core shared resource of the Masonic Cancer Center, University of Minnesota, and the NCATS. The design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication is solely the responsibility of the authors and does not necessarily represent the official views of the funders or sponsors.
J.P.K. has financial connections with Varian; H.A.S. has financial connections with Genentech; D.R. has financial connections with Varian, Siemens, Accuray, BrainLab, and Elekta; and M.P.M. has financial connections with Abbott, Novelos, Phillips, BMS, Celldex, Roche, Elekta, Novocure, Novartis, Cavion, and Pharmacyclics.
Footnotes
An online CME test for this article can be taken at https://academy.astro.org.
Conflict of interest: The other authors report no conflict of interest.
References
- 1.Gavrilovic IT, Posner JB. Brain metastases: Epidemiology and pathophysiology. J Neurooncol 2005;75:5–14. [DOI] [PubMed] [Google Scholar]
- 2.Hodi FS, O’Day SJ, McDermott DF, et al. Improved survival with ipilimumab in patients with metastatic melanoma. N Engl J Med 2010; 363:711–723 (Erratum, N Engl J Med 2010;363:1290). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Barnholtz-Sloan JS, Sloan AE, Davis FG, et al. Incidence proportions of brain metastases in patients diagnosed (1973–2001) in the Metropolitan Detroit Cancer Surveillance System. J Clin Oncol 2004;22: 2865–2872. [DOI] [PubMed] [Google Scholar]
- 4.Postow MA, Chesney J, Pavlick AC, et al. Nivolumab and ipilimumab versus ipilimumab in untreated melanoma. N Engl J Med 2015;372: 2006–2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Gaspar LE, Scott C, Rotman M, et al. Recursive partitioning analysis (RPA) of prognostic factors in three Radiation Therapy Oncology Group (RTOG) brain metastases trials. Int J Radiat Oncol Biol Phys 1997;37:745–751. [DOI] [PubMed] [Google Scholar]
- 6.Sperduto PW, Berkey B, Gaspar LE, et al. A new prognostic index and comparison to three other indices for patients with brain metastases: An analysis of 1960 patients in the RTOG database. Int J Radiai Oncol Biol Phys 2008;70:510–514. [DOI] [PubMed] [Google Scholar]
- 7.Sperduto PW, Chao ST, Sneed PK, et al. Diagnosis-specific prognostic factors, indexes, and treatment outcomes for patients with newly diagnosed brain metastases: A multi-institutional analysis of 4,259 patients. Int J Radiat Oncol Biol Phys 2010;77:655–661. [DOI] [PubMed] [Google Scholar]
- 8.Sperduto PW, Kased N, Roberge D, et al. The Graded Prognostic Assessment (GPA): A new diagnosis-specific prognostic index for women with breast cancer and brain metastases. Int J Radiat Oncol Biol Phys 2012;82:2111–2117. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Sperduto PW, Kased N, Roberge D, et al. Summary report on the Graded Prognostic Assessment (GPA): An accurate and facile diagnosis-specific tool to estimate survival, guide treatment and stratify clinical trials for patients with brain metastases. J Clin Oncol 2012;30:419–425. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Sperduto PW, Yang TJ, Beal K, et al. The effect of gene alterations and tyrosine kinase inhibition on survival and cause of death in patients with adenocarcinoma of the lung and brain metastases. Int J Radiat Oncol Biol Phys 2016;96:406–413. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Sperduto PW, Yang TK, Beal K, et al. Improved survival and prognostic ability in lung cancer patients with brain metastases: An update of the Graded Prognostic Assessment for lung cancer using molecular markers (Lung-molGPA). JAMA Oncol 2017;3:827–831. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Sperduto PW, Jiang W, Brown PD, et al. The prognostic value of BRAF, cKIT and NRAS mutations in melanoma patients with brain metastases. Int J Radiat Oncol Biol Phys 2017;98:1069–1077. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Sperduto PW, Hall WA. Radiosurgery, cost-effectiveness, gold standards, the scientific method, cavalier cowboys and the cost of hope. Int J Radiat Oncol Biol Phys 1996;36:511–513. [DOI] [PubMed] [Google Scholar]
- 14.Mulvenna P, Nankivell M, Barton R, et al. Dexamethasone and supportive care with or without whole brain radiotherapy in treating patients with non-small cell lung cancer with brain metastases unsuitable for resection or stereotactic radiotherapy (QUARTZ): Results from a phase 3, non-inferiority, randomised trial. Lancet 2016;388:2004–2014. [DOI] [PMC free article] [PubMed] [Google Scholar]