SEARCH

SEARCH BY CITATION

Keywords:

  • brain metastasis;
  • melanoma;
  • survival;
  • prognostic factors;
  • stereotactic radiosurgery;
  • neurosurgery;
  • serum lactate dehydrogenase

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

BACKGROUND:

This multicenter study aimed to identify prognostic factors in patients with brain metastases from malignant melanoma (BM-MM).

METHODS:

In a retrospective survey in 9 cancer centers of the German Cancer Society, 692 patients were identified with BM-MM during the period 1986 through 2007. Overall survival was analyzed using a Kaplan-Meier estimator and compared with log-rank analysis. Cox proportional hazards models were used to identify prognostic factors significant for survival.

RESULTS:

The median overall survival of the entire cohort was 5.0 months (95% confidence interval [95% CI], 4 months-5 months). Significant prognostic factors in the univariate Kaplan-Meier analysis were Karnofsky performance status (≥70% vs <70%; P < .001), number of BM-MM (single vs multiple; P < .001), pretreatment levels of lactate dehydrogenase (LDH) (normal vs elevated; P < .001) and S-100 (normal vs elevated; P < .001), prognostic groups according to Radiation Therapy Oncology Group (class I vs class II vs class III; P = .0485), and treatment choice (for the cohort with single BM-MM only) (stereotactic radiotherapy or neurosurgical metastasectomy vs others; P = .036). Cox proportional hazards models revealed pretreatment elevated level of serum LDH (hazard ratio [HR], 1.6; 95% CI, 1.3-2.0 [P = .00013]) and number of BM-MM (HR, 1.6; 95% CI, 1.3-2.0 [P = .00011]) to be independent prognostic variables in the entire cohort, whereas in patients with a single BM-MM, treatment choice (HR, 1.5; 95% CI, 1.1-1.9 [P = .0061]) was identified as a unique prognostic factor.

CONCLUSIONS:

The overall survival of patients with BM-MM primarily depends on the number of metastases and pretreatment level of LDH. In the case of a single brain metastasis, stereotactic radiotherapy or neurosurgical metastasectomy is by far the most important factor for improving survival. Cancer 2011. © 2010 American Cancer Society.

Metastatic spread to the brain is the most serious event in the course of melanoma because it carries the worst prognosis of all visceral metastases and represents the major cause of death in patients with disseminated disease.1 Large clinical series have demonstrated that brain metastases (BM) are diagnosed in up to 10% of melanoma patients during their lifetime and autopsy data have revealed that up to 73% of patients who died from disseminated cutaneous melanoma (CM) had BM.2-4

The recent management of cerebral metastases mainly depends on the number and size of the metastases and on the extracranial extension of the disease. It involves neurosurgery, stereotactic radiosurgery (STR), whole brain radiotherapy (WBRT), and chemotherapy. Surgical excision or radiosurgery/gamma knife are the first treatment choices for patients with solitary BM.5 STR is likewise offered as an alternative to neurosurgery in patients with a few BMs that cannot be surgically removed. WBRT is usually used in conjunction with neurosurgery or STR because it is associated with better local control. Patients who are not eligible for surgery or STR are usually offered WBRT, although to our knowledge only neurological symptoms are improved and no survival advantage has been demonstrated to date.6, 7 Patients with a poor Karnofsky performance status (KPS) may be managed with steroids and other supportive measures. To the best of our knowledge, no generally accepted standards for the combinations or sequence of these therapies exist to date.8, 9

The objective of this retrospective study was to analyze prognostic factors, effects of treatment, and survival outcome in relatively nonselected patients with melanoma metastatic to the brain to identify meaningful prognostic groups with respect to overall survival (OS).

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

Clinical and pathological data of patients diagnosed between 1986 and 2007 with BM from malignant melanoma (BM-MM) were obtained retrospectively from 9 cancer centers of the German Cancer Society. All data were recorded within standardized forms and finally computerized. The following information was obtained: demographics, anatomical localization and histological characteristics of the primary melanoma, time interval between diagnoses of primary melanoma and BM, site of primary CM, clinical presentation of BM, number and size of BM, number and location of other CM metastases, KPS, serum lactate dehydrogenase (LDH), serum protein S-100, administered treatment, response to treatment, date of death or last follow-up, and cause of death. Diagnosis of BM-MM was based on computed tomography (CT) and magnetic resonance imaging (MRI) scans. Patients were routinely screened by CT. In the case of suspicious findings, MRI was performed to validate the findings of the CT scan. Patients were followed according to current surveillance programs, including re-scans of the brain every 8 to 12 weeks by CT or MRI.

Statistical Analysis

Follow-up time was defined as the date of last follow-up or death minus the date of diagnosis of BM-MM (OS). Brain metastases diagnosed within 30 days of the diagnosis of the primary melanoma were considered synchronous. Survival was calculated from the date of diagnosis of BM-MM. For calculations concerning neurosurgery or STR, patients with a single BM-MM were considered only because the maximum number of BM-MM for which neurosurgery and/or STR was offered to the patients varied throughout the 9 centers. In the survival analyses, deaths whatever the cause were considered. Survival curves and median survival with relative 95% confidence intervals (95% CIs) were generated according to the Kaplan-Meier estimators and compared using log-rank tests. Variables proven significant in the univariate analyses at the .05 level were included in a Cox proportional hazards model. Categorical variables were dummy-coded and P values were based on the Wald test. Throughout the analysis, P values <.05 were considered statistically significant. All statistical computations were performed using R 2.6 by the R Foundation of Statistical Computing (Vienna, Austria).

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

A total of 692 medical record files were recorded from patients treated between 1986 and 2007. Follow-up data were available for 672 of those 692 patients. The clinical characteristics of all patients are summarized in Table 1. The median time from the diagnosis of the primary melanoma until the development of BM-MM was 30.4 months. In 22.8% of patients, BM-MM developed during the first year after the initial diagnosis of distant metastases, including 62 patients in whom BM were diagnosed at the time of first presentation. In nearly all patients who died, the cause of death was progressive disease. One patient committed suicide. The median OS of the entire cohort was 5.0 months (95% CI, 4 months-5 months) after a diagnosis of BM-MM. Prognostic factors found to be significant in the univariate Kaplan-Meier analysis were KPS (≥70% vs <70%; P < .001), number of BM-MM (single vs multiple; P < .001), pretreatment levels of serum LDH (normal vs elevated; P < .001), pretreatment levels of serum protein S-100B (normal vs elevated; P < .001), prognostic groups according to Radiation Therapy Oncology Group (RTOG) (class I vs class II vs class III; P = .0485), response to treatment (progressive disease vs stable disease vs partial response vs complete response; P < .001), and type of applied treatment (for the cohort with single BM-MM only) (stereotactic radiotherapy or neurosurgical metastasectomy vs others; P = .036) (Table 2) (Fig. 1). In patients who were treated with surgery or STR (n = 200 patients), the addition of WBRT was found to prolong OS (9 months [95% CI, 8 months-14 months] vs 7 months [95% CI, 6 months-9 months]; P = .0458). Nonstatistically significant factors included gender (median survival for males, 4 months [95% CI, 3 months-5 months] vs 5 months for females [95% CI, 4 months-6 months]; P = .117), the presence of extracerebral metastases (median survival for present extracerebral metastases, 4 months [95% CI, 4 months-5 months] vs 6 months [95% CI, 5 months-7 months] for absence of extracerebral metastases; P = .0562), time frame between the initial diagnoses of MM and the development of BM-MM (P = .110), and year of diagnosis of BM (1986-1990, 1991-1995, 1996-2000, 2001-2005, and 2006-2007; P = .258). Patients treated with temozolomide or fotemustine (n = 163) did not have a favorable OS compared with patients (n = 179) treated with other chemotherapeutic schedules as first-line treatment (5 months [95% CI, 4 months-5 months] vs 6 months [95% CI, 5 months-7 months]; P = .298).

thumbnail image

Figure 1. Kaplan-Meier plots are shown for (A) Karnofsky performance status, (B) number of brain metastases from malignant melanoma (BM-MM), (C) serum protein S-100 level, (D) serum lactate dehydrogenase level, (E) prognostic groups according to the Radiation Therapy Oncology Group, and (F) applied treatment (cohort with single BM-MM only). STR indicates stereotactic radiosurgery.

Download figure to PowerPoint

Table 1. Characteristics of 692 Patients
VariableValue
  • BM indicates brain metastases.

  • a

    Information available from 510 patients.

Age, y
 Median59
 Range7-100
Gender
 Male377 (54.5%)
 Female301 (43.5%)
 Missing data14 (2.0%)
Time from initial diagnosis of melanoma, y
 ≤1158 (22.8%)
 1-2128 (18.5%)
 2-3119 (17.2%)
 3-485 (12.3%)
 4-532 (4.6%)
 5-10119 (17.2%)
 ≥1051 (7.4%)
No. of BM
 1249 (36.0%)
 2102 (14.7%)
 353 (7.7%)
 >3242 (35.0%)
 Meningiosis12 (1.7%)
 Missing data34 (4.9%)
Sum of largest diameters, mma
 Median36
 Range2-70
Extracerebral visceral metastases
 None147 (21.2%)
 1 organ only200 (28.9%)
 2–3 organs involved271 (39.2%)
 ≥4 organs affected74 (10.7%)
Distribution of extracranial metastases
 Lung439 (63.4%)
 Liver260 (37.6%)
 Bones110 (15.9%)
 Skin275 (39.7%)
 Lymph nodes430 (62.1%)
 Others212 (30.6%)
Table 2. Kaplan-Meier Estimators for Different Prognostic Factorsa
FactorNo. of Patients (%)Median Survival, Months (95% CI)P
  • 95% CI indicates 95% confidence interval; BM, brain metastases; KPS, Karnofsky performance status; LDH, lactate dehydrogenase; RTOG, Radiation Therapy Oncology Group; STR, stereotactic radiosurgery; WBRT, whole brain radiotherapy.

  • a

    Survival data were available from 672 patients.

  • b

    Including those patients in whom BMs were diagnosed at the time of first presentation. The median survival for those patients was 5 months (95% CI, 3 months-7 months).

Entire cohort672 (100.0)5 (4-5) 
Gender
 Male369 (56.0)4 (3-5).117
 Female290 (44.0)5 (4-6)
Time from initial diagnosis of melanoma, y
  ≤1b156 (23.2)4 (3-5).110
  1-2122 (18.2)4 (3-5)
  2-3115 (17.1)5 (4-7)
  3-482 (12.2)5 (3-7)
  4-532 (4.8)6 (4-9)
  5-10115 (17.1)5 (4-6)
  >1050 (7.4)7 (5-10)
5-y interval of diagnosis of BM
 1986-199016 (2.4)4 (2-7).258
 1990-199557 (8.5)6 (5-9)
 1996-2000204 (30.4)4 (3-5)
 2001-2005349 (51.9)5 (4-5)
 2006-200746 (6.8)4 (3-7)
KPS, %
 ≥70355 (66.5)6 (5-6)<.001
 <70179 (33.5)3 (3-4)
No. of BM
 Single240 (38.3)7 (6-9)<.001
 Multiple387 (61.7)4 (3-4)
LDH (at time point of BM diagnosis)
 Not elevated206 (44.4)6 (5-7)<.001
 Elevated258 (55.6)3 (3-4)
S-100 (at time point of BM diagnosis)
 Not elevated55 (20.4)7 (6-10)<.001
 Elevated215 (79.6)5 (4-5)
Extracerebral metastases
 Not present146 (21.7)6 (5-7).0562
 Present526 (78.3)4 (4-5)
RTOG prognostic groups
 Class I59 (11.1)7 (5-9).0485
 Class II374 (70.2)5 (4-5)
 Class III100 (18.8)3 (3-5)
Response to treatment
 Progressive disease373 (72.3)4 (3-5)<.001
 Stable disease68 (13.2)9 (7-11)
 Partial response42 (8.1)9 (8-13)
 Complete response33 (6.4)13 (10-18)
Type of treatment for single brain metastases
 STR or surgery (complete resection)122 (57.0)9 (7-11).036
 WBRT and/or chemotherapy92 (43.0)6 (5-7)
Type of treatment for limited brain disease (<3 metastases)
  STR121 (67.2)6 (5-7).448
  WBRT59 (32.8)7 (6-11)

Cox proportional hazards models revealed that an elevated pretreatment level of serum LDH (hazard ratio [HR], 1.6; 95% CI, 1.3-2.0 [P = .00013]) and number of BM-MM (HR, 1.6; 95% CI, 1.3-2.0 [P = .00011]) in the entire cohort of patients were independent prognostic variables (151 patients with available S-100 values at the time of diagnosis of BM [Table 3] and 354 patients independent of available S-100 values [Table 4]), whereas in patients with a single BM-MM, the type of applied treatment (STR or neurosurgical metastasectomy vs others: HR, 1.5; 95% CI, 1.1-1.9 [P = .0061]) was identified as a unique prognostic factor (Table 5).

Table 3. Multivariate Analysis of Predictive Prognostic Factors Including S-100 Levela,b
FactorHR (95% CI)P
  • HR indicates hazard ratio; 95% CI, 95% confidence interval; KPS, Karnofsky performance status; BM, brain metastases; LDH, lactate dehydrogenase.

  • a

    Radiation Therapy Oncology Group prognostic groups were not considered in this model because class I only accounted for 9 patients.

  • b

    Full data were available from 151 patients.

KPS, %
 ≥70Reference.04
 <702.7 (1.1-7.0)
No. of BM
 SingleReference.002
 Multiple1.6 (1.3-2.7)
LDH (at time point of BM diagnosis)
 Not elevatedReference.01
 Elevated1.6 (1.1-2.4)
S-100 (at time point of BM diagnosis)
 Not elevatedReference.75
 Elevated1.1 (0.7-1.7)
Table 4. Multivariate Analysis of Prognostic Factors Excluding S-100 Levela
FactorHR (95% CI)P
  • HR indicates hazard ratio; 95% CI, 95% confidence interval; KPS, Karnofsky performance status; BM, brain metastases; LDH, lactate dehydrogenase; RTOG, Radiation Therapy Oncology.

  • a

    Full data were available from 354 patients.

KPS, %
 ≥70Reference.17
 <701.2 (0.9-1.7)
No. of BM
 SingleReference.00011
 Multiple1.6 (1.3-2.0)
LDH (at time point of BM diagnosis)
 Not elevatedReference.00013
 Elevated1.6 (1.3-2.0)
RTOG prognostic groups
 Class IReference.80
 Class II/III1.1 (0.7-1.5)
Table 5. Multivariate Analysis of Prognostic Factors for Single BM Excluding S-100 Level (n=236)
FactorHR (95% CI)P
  1. BM indicates brain metastases; HR, hazard ratio; 95% CI, 95% confidence interval; KPS, Karnofsky performance status; LDH, lactate dehydrogenase; STR, stereotactic radiosurgery; WBRT, whole brain radiotherapy.

KPS, %
 ≥70Reference.83
 <701.1 (0.7-1.7)
LDH (at time point of BM diagnosis)
 Not elevatedReference.30
 Elevated1.2 (0.9-1.6)
Type of treatment
 STR or surgery (complete resection)Reference.0061
 WBRT and/or chemotherapy1.5 (1.1-1.9)

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

Melanoma is 1 of the most frequent cancer entities causing cerebral metastases leading to neurological complications.10 The purpose of the current analysis was to identify prognostic factors in regard to the outcome of patients diagnosed with BM-MM. Similar to previous studies,11, 12 the current evaluation was also retrospective. Patient characteristics in this analysis were comparable to those included in previous published studies, with a slight predominance of male patients.4, 12, 13 The median time from diagnosis of the primary melanoma to the development of BM-MM was 30.4 months, which is in keeping with data from the study by Choi et al14 but shorter when compared with analyses by Fife et al13 and Sampson et al,4 which reported a time frame of 3.1 and 3.7 years. In our cohort, the OS was poor, with a median survival time ≤5 months, which was, however, comparable to data reported by other investigators.13, 15, 16 It is interesting to note that the median survival was not different between the patients diagnosed with BM between 1986 and 1990, 1991 and 1995, 1996 and 2000, 2001 and 2005, and 2006 and 2007.

Patients diagnosed with a single cerebral lesion were found to have a more favorable prognosis than patients with multiple lesions or leptomeningeal disease (median survival of 7 months vs 4 months). These data are in keeping with data from Raizer et al.12 In contrast to other reports,12, 17 we were not able to detect a significant difference in OS between patients with extracerebral metastases and patients with BM-MM only.

In 1997, the RTOG proposed 3 prognostic groups based on recursive partitioning analysis (RPA) of their trials of radiotherapy for cerebral metastases.18 The median survival ranged from 7.1 months for RPA class I patients (controlled brain disease and without extracranial involvement) to 4.2 months for RPA class II patients (KPS >70%, uncontrolled brain disease, extracerebral disease, and/or age <65 years) and 2.3 months for RPA class III patients (KPS <70%;). It is interesting to note that, in our cohort of patients, the grouping according to the RTOG proposal was only significant in the univariate analysis but failed to achieve significance in the multivariate analysis. Similar results apply for KPS. Although a KPS <70% was a prognostic factor in the univariate analysis, it failed to achieve significance in the multivariate Cox regression model.

An increased level of LDH is an established prognostic factor in the classification of stage IV melanoma, resulting in an upstaging of M1a or M1b disease to M1c.19, 20 However, the relevance in patients with BM-MM has to our knowledge remained ambiguous to date. Data from the current study indicate a high prognostic value of LDH in patients with BM-MM validated by univariate and multivariate analyses. A study by Staudt et al recently confirmed these findings in a smaller cohort of patients.21 Although serum S-100 is a verifiable marker in the surveillance of patients with surgically resected, high-risk melanoma,22, 23 even surpassing the prognostic impact of LDH, its value in patients with BM-MM appears to be alleviated. In detail, serum protein S-100 was found to be a significant prognostic factor in the univariate analysis but failed in the multivariate Cox model.

The significance of STR in the management of patients with BM-MM has been demonstrated in various studies, achieving acceptable disease control rates.11, 12, 24, 25 Raizer et al12 and Seung et al25 both described median survival rates between 7 to 9 months that are comparable to our findings. Grob et al even reported a median OS of 22 months in patients with solitary BM-MM who were treated with STR, although in a small cohort of patients.11 In our group of patients with a single BM-MM, neurosurgery and STR were both found to be associated with a significantly longer OS compared with other treatment modalities such as WBRT and/or systemic therapy. These findings are in keeping with recent retrospective and randomized studies.12, 13, 26, 27 However, this benefit is no longer detectable when considering patients with limited disease (<3 metastases) who are treated with STR versus WBRT. This emphasized the limitation of surgery and STR. Both methods are only available for patients with limited disease, who have a better prognosis than patients with multiple metastases per se.

An additional benefit of WBRT in conjunction with STR or surgery is still controversial. One investigation has demonstrated that WBRT may delay the development of new BM,17, 28 whereas others have not.24, 29 In our cohort of 200 evaluable patients who were treated with surgery or STR, the subsequent application of WBRT prolonged the median OS by 2 months (7 months vs 9 months). Not surprisingly, patients who responded to treatment had a better prognosis compared with those who failed treatment.

In the current study, we did not adjust the P values for multiple testing according to Bonferroni.30 However, a conservative approach with an α of .05 per 8 tests results in a modified P value of .00625. Therefore KPS, pretreatment LDH and S-100 levels, and number of BM retain their significant impact in the univariate analyses.

In conclusion, the OS of patients with BM-MM mainly depends on the number of metastases and pretreatment level of LDH. In the case of a single brain metastasis, the application of STR or neurosurgical metastasectomy is by far the most important factor for improving survival. In contrast to the RTOG, we suggest not only to include the number of BM, patient age, and KPS in the treatment planning but also pretreatment levels of S-100 and LDH. Because the findings of the current analysis are based on retrospective data, we recommend confirming the evaluation in a prospective setting.

CONFLICT OF INTEREST DISCLOSURES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

Supported in part by an educational grant from Essex/Schering-Plough.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES