Malignant peripheral nerve sheath tumors

Prognostic factors and survival in a series of patients treated at a single institution

Authors


Abstract

BACKGROUND.

The authors explored the prognostic factors and clinical outcomes of patients who had malignant peripheral nerve sheath tumors (MPNST) with and without neurofibromatosis type 1 (NF-1).

METHODS.

Two hundred five patients with localized MPNST who underwent surgery at the Istituto Nazionale per lo Studio e la Cura dei Tumori (Milan, Italy) over 25 years were reviewed. Forty-six patients had concomitant NF-1 syndrome, and 159 patients did not. Local recurrence, distant metastases, and survival rates were studied.

RESULTS.

One hundred thirty patients presented with primary disease, and 75 patients had locally recurrent tumors. The disease-specific mortality rate was 43% at 10 years, with a continuously disease-free survival rate of no greater than 40%. Presentation with either primary or recurrent disease, tumor size, and tumor site (trunk vs. extremity) were the strongest independent predictors of survival. Margin status and radiation therapy also played a role, mostly related to their effect on local outcome. Pathologic grade influenced distant metastases, but only a trend for survival could be observed. No significant independent differences between patients with and without NF-1 were observed.

CONCLUSIONS.

To the authors' knowledge, this was among the largest single-institution series to date. The results confirmed that patients with MPNST share similar prognostic factors with patients who have other soft tissue sarcomas and have some of the worst clinical outcomes. The presence of NF-1 syndrome per se did not affect survival, but patients with NF-1 were more likely to have larger tumors. Therefore, such patients should be followed carefully to detect disease as early as possible. Cancer 2006. © 2006 American Cancer Society.

Malignant peripheral nerve sheath tumors (MPNSTs) are rare, with an expected incidence of 0.1/100,000 per year.1 They are defined as any malignant tumor arising from or differentiating toward cells of the peripheral nerve sheath.

MPNSTs account for approximately 5% to 10% of all soft tissue tumors.1 They have a known association with neurofibromatosis type 1 (NF-1), an autosomal-dominant disorder that involves the NF1 tumor suppressor gene, which is located on chromosome 17. The reported incidence of MPNST in this subgroup of patients varies widely in the literature, ranging from 2% to 29%.2, 3 Recently, it was reported that this life-time risk was associated with the type of gene mutation. In fact, most patient with MPNSTs display a small, truncating mutation in the NF1 gene that is associated with an 8% to 13% risk of developing MPNST. However, from 5% to 10% of individuals with NF-1 have a microdeletion that incorporates the entire NF1 gene and a consequent risk for MPNST of from 16% to 26%.4–7

Approximately 25% to 50% of observed MPNSTs occur in patients with NF-1. Like other soft tissue sarcomas, MPNSTs have a tendency to recur locally and spread hematogenously. Despite aggressive surgery and adjuvant therapy, the prognosis for patients with MPNST remains poor; however, to date, prognostic factors have not been identified consistently in the literature.8–16 Therefore. we analyzed a series of patients with MPNST who were treated consecutively at our institution from 1976 to 2003 to gain a better understanding of the natural history of the disease and to investigate its prognostic factors, with particular attention to the factors that affect survival.

MATERIALS AND METHODS

Between January 1976 and December 2003, 221 consecutive patients with MPNST underwent surgery with the intent to eradicate disease at the Istituto Nazionale per lo Studio e la Cura dei Tumori (Milan, Italy). From this series, we excluded 16 patients who presented with synchronous metastases. Thus, our series includes 205 patients who had localized disease.

An attempt was made to perform a complete resection with negative margins in all patients. In extremities, the surgical procedure was similar to what normally was performed for all other soft tissue sarcomas. In the trunk and head and neck, wide margins only rarely were achievable.

Each patient was assessed clinically for the presence of NF-1 syndrome. Thus, patients were assigned to the NF-1 subgroup if at least 2 of the following criteria were present: ≥6 cafe-au-lait macules (>5 mm before puberty, >15 mm after puberty), skin-fold freckles (groin, axilla, neck base), ≥2 neurofibromas (1 plexiform), skeletal dysplasia (orbital or tibial), Lisch nodules (≥2 iris hamartomas), optic glioma, and family history.

Outside of the context of NF-1 or involvement of a nerve, the pathologic diagnosis of MPNST is a subject of some considerable controversy. Nevertheless, during the study period, histologic diagnoses were reviewed prospectively by ≥2 experienced pathologists at our institution. The morphologic criteria to confirm the diagnosis were evidence of dense fascicles of spindle cells with frequent occurrence of necrosis alternating with less cellular areas in a marbleized appearance. Perivascular accentuation of cellularity always was observed. The expression of S100 protein was rare, whereas nerve growth factor receptor expression often was present. Cytokeratins and epithelial membrane antigen were absent. Muscolar markers generally were absent except in tumors that displayed rhabdomioblastic differentiation (Triton tumors).

The French Federation of Cancers Centers Against Cancer (FNCLCC) grading system was applied for tumor grade assessment.17 In patients with recurrent tumors, this assessment was performed on the slides from the primary tumor that was seen in consultation. In 17 patients, tumor grade could not be assessed.

The surgical specimen always was examined in the presence of the operating surgeon. The margins were inked and sampled separately. Margins were classified as positive when the resection was intralesional or marginal (closest margin within 1 mm from the inked surface) or negative otherwise.

The indication for radiation therapy was given by both the operating surgeon and the radiation oncologist when there was a supposed greater risk of recurrence on clinical grounds. However, no prospectively selected criteria were used to this end. Each patient was discussed within our multidisciplinary Soft Tissue Sarcoma Group. External beam radiation was given to all such patients with that doses ranged from 45 grays (Gy) to 65 Gy (median, 57 Gy).

Chemotherapy was given at the discretion of the multidisciplinary Soft Tissue Sarcoma Group of our institution or as part of clinical trials. Anthracycline-based regimens were used and, in most patients, were associated with ifosfamide. Given the time span of our case series, no uniform criteria for both radiation therapy and chemotherapy were applied.

Statistical Methods

Two distinct, multivariate, binary logistic models were used to test the homogeneity between patients with and without NF-1 syndrome and between patients with primary or recurrent tumors with respect to other patient and tumor characteristics. The binary response variable took the value of 1 for patients with NF-1 in the first model or patients with recurrent tumors in the second model, and it took the value of 0 for the complementary groups of patients. Patient age and tumor size were modeled as continuous variables by using 3-knot, restricted cubic splines, whereas the other variables were modeled as categorical by using dummy (0 or 1) variables.18 The effect of each characteristic was tested by the corresponding 2-sided Wald test.

The endpoints of this study were cause-specific mortality, local recurrence, and distant metastasis. The time to occurrence of any event was computed from the date of surgery at our institution to the date when the event first was recorded, or the event was censored at the date of last follow-up assessment in event-free patients.

Crude cumulative incidence curves for each endpoint were calculated in a competing-risks framework.19 In the analysis of cause-specific mortality, deaths caused by conditions unrelated to sarcoma were regarded as competing events. For the local recurrence (distant metastasis) analysis, deaths without evidence of disease and distant metastasis (local recurrence), whichever occurred first, were regarded as competing events. Concomitant local recurrences and distant metastases were included in the estimation of the crude cumulative incidence curves as distant metastases.

Multivariate analyses of each endpoint were based on cause-specific hazards and, thus, were performed using Cox multiple regression models. The following covariates were included in the models: age, presentation (primary or recurrent), syndrome (NF-1 or no NF-1), tumor site, tumor size, tumor grade (Grade 1, 2, or 3), margin status (negative or positive), chemotherapy (yes or no), and radiotherapy (yes or no). Age was modeled as continuous variables by using 3-knot, restricted cubic splines, whereas a linear term was used for size, because a linear relation between the relative hazard and size was evident for all 3 endpoints investigated.18 All of the other covariates were modeled as categorical by using dummy (0 or 1) variables. Checking the proportional hazard assumption implied by the Cox model relied on graphic analyses of scaled Schoenfeld residuals.20 To test whether the prognostic effect of age differed among patients with different syndromes, the interaction terms age × syndrome were inserted into the multiple Cox model. These terms were not significant statistically for all endpoints investigated; thus, we have reported the results from the models without interactions.

The low number of observed events prevented us from performing analyses with reliable multivariate Cox models in the 2 subsets of primary and recurrent tumors. However, we compared the multivariate Cox models fitted on each of the 2 subsets of primary and recurrent tumors with the model obtained on the overall case series. Each comparison was carried out by performing a likelihood ratio test of the model fitted on the subset versus the model in which the coefficients were held fixed and equal to the coefficients estimated on the overall case series. In both models, the test results were not statistically significant. Thus, we have reported only the results from the models that were performed on the overall case series.

We used SAS™ software (SAS/STAT Software: Changes and Enhancements through Release 6.11; SAS Institute Inc., Cary, NC) and the S-Plus® (StatSci; MathSoft, Seattle, WA) Design and Cmprsk (available at URL: http://biowww.dfci.harvard.edu/˜gray/) libraries to perform the modeling and statistical calculations. Two-sided P values below the 5% conventional threshold were considered significant.

RESULTS

The main characteristics of 205 study patients are summarized in Table 1 either overall or according to the presence of NF-1 syndrome. The median age at presentation for the whole series was 37 years (range, 1–90 years; interquartile range, 24–57 years), and there was a slight male predominance (52.7%). Forty-six patients (22.4%) developed the disease associated with NF-1 syndrome, and the other 159 patients (77.6%) had no features of NF-1 syndrome. One hundred thirty patients (63.4%) presented with primary tumors, and 75 patients (36.6%) presented with locoregional, recurrent tumors. Tumors were located in the trunk in 50.7% of patients, in the extremities in 44.9% of patients, and in the head and neck in 4.4% of patients. Negative margins were achieved with surgery in 127 patients (62.0%). Fifty-three patients (25.9%) received radiation therapy, 24 patients (11.7%) received chemotherapy, and 38 patients (18.5%) received both.

Table 1. Main Patient and Disease Characteristics According to Neurofibromatosis Type 1 Syndrome
CharacteristicWith NF-1Without NF-1Overall
No.%No.%No.%
  1. NF-1 indicates neurofibromatosis type 1; IQ, interquartile; FNCLCC, French Federation of Cancers Centers.

Total46100159100205100
Gender
 Female235074479747
 Male2350855310853
Median age (IQ range), y27(18–37)40(26–59)37(24–57)
Presentation
 Primary3372976113063
 Recurrent132862397537
Tumor site
 Head and neck128594
 Trunk2350815110451
 Extremity224870449245
Median tumor size (IQ range), cm
Depth10.0(8.0–15.0)6.0(3.0–10.0)7.0(4.0–12.0)
 Superficial2350996212260
 Deep235060388340
FNCLCC grade
 Grade 112138147
 Grade 261324153015
 Grade 334741106914470
 Missing511128178
Margin status
 Negative2963986212762
 Positive173761387838
Radiation therapy
 Not done2554895611456
 Done214670449144
Chemotherapy
 Not done23501207514370
 Done235039256230

Based on the logistic model, no significant differences between patients with and without NF-1 syndrome were identified according to gender, tumor site, depth, tumor grade, tumor margins, and radiation therapy administration. The 2 groups differed significantly with regard to patient age (P = .002), tumor size (P = .001), and chemotherapy administration (P = .033). In particular, on average, patients with NF-1 syndrome were younger and had larger tumors compared with patients without NF-1 syndrome (median age, 27 years vs. 40 years, respectively; median tumor size, 10 cm vs. 6 cm, respectively; see Table 1).

No significant differences between patients with primary and recurrent tumors were observed in the logistic model analysis according to gender, tumor site, tumor size, depth, grading, tumor margins, chemotherapy, or radiation therapy. Patients with primary and recurrent disease differed significantly with regard to age (P = .008) and the presence of NF-1 syndrome (P = .012). In particular, compared with patients who had recurrent tumors, patients who had primary tumors were younger (median age, 32 years vs. 40 years) and more frequently had NF-1 syndrome (25.4% vs. 17.3%).

The median follow-up for the entire group was 112 months (interquartile range, 54–160 months) in September 2005. For patients with primary tumors, the median follow-up was 110 months (interquartile range, 53–159 months); and, for patients who presented with recurrent tumors, it was 119 months (interquartile range, 28–160 months). The 2-year, 5-year and 10-year cumulative incidence estimates for the 3 endpoints investigated are shown in Table 2.

Table 2. Crude Cumulative Incidence Estimates for the 3 Endpoints Investigated*
 Cause-specific mortalityLocal recurrenceDistant metastasis
2-Year5-Year10-Year2-Year5-Year10-Year2-Year5-Year10-Year
  • NF-1 indicates neurofibromatosis type 1; FNCLCC, French Federation of Cancers Centers.

  • *

    The crude cumulative incidence (%) is shown with the standard error in parenthesis.

Overall26.6 (3.4)39.9 (3.7)43.3 (3.8)23.3 (2.9)27.3 (3.2)28.8 (3.3)22.8 (3.0)26.2 (3.1)28.7 (3.3)
Presentation
 Primary21.8 (3.7)36.9 (4.6)39.8 (4.8)24.0 (3.8)26.7 (3.9)26.7 (3.9)17.9 (3.4)21.5 (3.7)25.6 (4.2)
 Recurrent35.2 (5.7)45.2 (6.2)49.2 (6.4)22.1 (4.9)28.5 (5.5)32.1 (5.8)31.5 (5.5)34.5 (5.7)34.5 (5.7)
Syndrome
 Non-NF-123.5 (3.4)38.9 (4.2)39.8 (4.2)21.7 (3.3)24.7 (3.5)26.6 (3.7)19.9 (3.2)24.1 (3.5)27.3 (3.8)
 NF-138.4 (7.7)43.9 (8.0)54.4 (8.6)29.2 (6.9)37.1 (7.5)37.1 (7.5)33.5 (7.2)33.5 (7.2)33.5 (7.2)
Tumor site
 Head and neck44.4 (17.9)66.7 (17.6)66.7 (17.6)22.2 (15.1)22.2 (15.1)22.2 (15.1)33.3 (17.4)33.3 (17.4)33.3 (17.4)
 Trunk37.2 (5.0)48.8 (5.4)52.9 (5.7)30.5 (4.6)35.2 (4.8)35.2 (4.8)24.6 (4.4)27.7 (4.5)27.7 (4.5)
 Extremity13.4 (3.6)27.5 (5.0)30.7 (5.3)15.4 (3.8)19.2 (4.2)22.2 (4.6)17.5 (4.0)23.7 (4.6)28.6 (5.1)
Tumor size
 <5 cm6.3 (2.7)23.7 (5.1)25.7 (5.4)14.7 (4.0)18.7 (4.4)22.3 (4.9)12.1 (3.6)20.0 (4.5)21.8 (4.5)
 >5 cm to <10 cm32.1 (6.1)47.0 (7.0)49.8 (7.2)27.9 (5.9)31.7 (6.1)31.7 (6.1)25.2 (5.7)25.2 (5.7)28.7 (6.5)
 >10 cm50.1 (6.8)56.3 (6.9)61.6 (7.1)30.3 (6.0)34.6 (6.4)34.6 (6.4)35.2 (6.2)35.2 (6.2)37.6 (6.5)
FNCLCC grade
 Grade 17.7 (7.7)7.7 (7.7)7.7 (7.7)7.7 (7.8)7.7 (7.8)7.7 (7.8)7.1 (7.1)7.1 (7.1)7.1 (7.1)
 Grade 217.4 (7.1)31.1 (9.7)39.2 (11.8)20.3 (7.8)29.7 (9.4)29.7 (9.4)16.8 (7.0)20.7 (7.8)20.8 (7.8)
 Grade 330.3 (4.0)44.0 (4.5)47.2 (4.6)24.1 (3.6)27.3 (3.8)29.2 (3.9)25.4 (3.7)29.5 (3.9)32.5 (4.1)
Margin status
 Negative15.7 (3.3)28.2 (4.3)33.6 (4.8)13.6 (3.1)20.2 (3.7)21.4 (3.9)21.7 (3.7)25.4 (4.0)27.2 (4.3)
 Positive44.7 (5.9)58.9 (6.0)58.9 (6.0)38.9 (5.6)38.9 (5.6)40.5 (5.7)24.7 (4.9)27.7 (5.2)31.0 (5.5)
Radiation therapy
 Yes25.8 (4.8)38.3 (5.5)40.0 (5.7)22.5 (4.5)29.0 (4.4)26.8 (4.9)22.4 (4.4)28.9 (4.9)32.2 (5.3)
 No27.4 (4.3)41.2 (4.9)45.8 (5.2)24.0 (4.0)25.2 (4.7)30.3 (4.5)23.2 (4.0)24.2 (4.1)26.0 (4.4)

Disease-Specific Mortality

Of the 205 investigated patients, 93 deaths were recorded. Of these, 79 deaths were related to disease and, thus, were considered in the current analyses. There were 46 cause-specific deaths among 130 patients with primary tumors and 33 cause-specific deaths among 75 patients with recurrent tumors.

The Cox multivariate regression model (Table 3) yielded significant results for presentation, tumor site, tumor size, margin status, and radiotherapy. In particular, the unfavorable factors were recurrent disease, head and neck or trunk tumor location, large tumor size, positive margins, and lack of radiotherapy. In particular, the larger the tumor, the greater the risk, which roughly was doubled (hazard ratio, 1.9) for an increase in tumor size from 4 cm to 12 cm. Young age, the presence of NF-1 syndrome, and Grade 3, although they were not significant statistically, had an unfavorable prognostic trend. Figure 1 shows the crude cumulative curves according to the presence or absence of NF-1 syndrome.

Figure 1.

This graph illustrates crude cumulative incidence curves of cause-specific death for patients with neurofibromatosis type 1 (NF-1) syndrome and patients without NF-1 syndrome (non-NF-1).

Table 3. Hazard Ratio Estimates with 95% Confidence Intervals and P Values from the Cox Proportional Hazards Model for Cause-Specific Survival
Category (Reference)HR95% CIP
  • HR indicates hazard ratio; 95% CI, 95% confidence interval; NF-1, neurofibromatosis type 1.

  • *

    Values correspond to the 75th and 25th percentiles of the variable distribution.

  • Overall 2-sided Wald test.

Age: 57 y (24 y)*0.840.54–1.33.248
Presentation: recurrence (primary)2.581.51–4.39<.001
Syndrome: NF-1 (no NF-1)1.370.76–2.47.303
Tumor site: head and neck (extremity)4.151.13–15.27.004
Tumor site: trunk (extremity)2.541.42–4.52 
Tumor size: 12 cm (4 cm)1.901.50–2.39<.001
Grade 3 (Grade 1 or 2)1.830.91–3.71.091
Margin status: positive (negative)1.861.10–3.16.021
Radiation therapy: yes (no)0.530.31–0.89.016
Chemotherapy: yes (no)1.650.95–2.86.073

Local Recurrence

Seventy-eight patients had local recurrences after they underwent surgery at our institute, including 43 patients with primary disease and 35 patients with local recurrences at presentation. Local recurrence as a first event occurred in 34 patients with primary disease and in 23 patients with recurrence at presentation. Sixty-four patients had only 1 local recurrence, and 14 patients had ≥2 local recurrences.

The time to local recurrence varied from 2 months to 159 months, with a median time to recurrence of 9 months. Twenty-six patients had local recurrences only, whereas 52 patients had both local recurrences and distant metastases.

The Cox multivariate regression model (Table 4) yielded significant results for presentation, tumor site, tumor size, and margin status. Even in this model, young age, the presence of NF-1 syndrome, and Grade 3 tumors, although they were not significant statistically, had an unfavorable prognostic trend. In particular, the larger the tumor, the greater the risk, which increased by 40% (hazard ratio, 1.4) for a variation in tumor size from 4 cm to 12 cm. Figure 2 shows the crude cumulative curves according to the presence or absence of NF-1 syndrome.

Figure 2.

This graph illustrates crude cumulative incidence curves of local recurrence for patients with neurofibromatosis type 1 (NF-1) syndrome and patients without NF-1 syndrome (non-NF-1).

Table 4. Hazards Ratio Estimates with 95% Confidence Intervals and P Values from the Cox Proportional Hazards Model for Local Recurrence
Category (Reference)HR95% CIP
  • HR indicates hazard ratio; 95% CI, 95% confidence interval; NF-1, neurofibromatosis type 1.

  • *

    Values correspond to the 75th and 25th percentiles of the variable distribution.

  • Overall 2-sided Wald test.

Age: 57 y (24 y)*0.830.53–1.31.290
Presentation: recurrence (primary)2.991.75–5.09<.001
Syndrome: NF-1 (no NF-1)1.740.94–3.23.077
Tumor site: head and neck (extremity)1.820.40–8.20.001
Tumor site: trunk (extremity)3.001.69–5.34 
Tumor size: 12 cm (4 cm)1.381.06–1.79.016
Grade 3 (Grade1 or 2I)1.660.87–3.57.118
Margin status: positive (negative)2.431.42–4.16.001
Radiation therapy: yes (no)0.680.41–1.14.146
Chemotherapy: yes (no)0.870.48–1.56.636

Distant Metastases

Eighty-seven patients had distant metastases after they underwent surgery at our center, including 48 patients with primary disease and 39 patients with local recurrence at presentation. Distant metastases as a first event occurred in 30 patients with primary disease and in 25 patients with recurrence at presentation.

Sixty-two patients had pulmonary metastasis only, whereas 23 patients also developed extrapulmonary metastases, and 2 patients had extrapulmonary metastases only. The sites of extrapulmonary metastases were the bone in 12 patients, the liver in 4 patients, the lymph nodes in 3 patients, and multiple sites (including soft tissue and the brain) in 6 patients. The time to distant metastasis varied from 1 month to 177 months, and the median time for those who developed distant metastases was 13 months.

The Cox multivariate regression model (Table 5) yielded significant results for presentation, tumor size, and tumor grade. In particular, the larger the tumor, the greater the risk, which increased by 70% (hazard ratio, 1.7) for a variation in tumor size from 4 cm to 12 cm. Figure 3 shows the crude cumulative curves according to the presence or absence of NF-1 syndrome.

Figure 3.

This graph illustrates crude cumulative incidence curves of distant metastases for patients with neurofibromatosis type 1 (NF-1) syndrome and patients without NF-1 syndrome (non-NF-1).

Table 5. Hazard Ratio Estimates with 95% Confidence Intervals and P Values from the Cox Proportional Hazard Model for Distant Metastases
Category (Reference)HR95% CIP
  • HR indicates hazard ratio; 95% CI, 95% confidence interval; NF-1, neurofibromatosis type 1.

  • *

    Values correspond to the 75th and 25th percentiles of the variable distribution.

  • Overall 2-sided Wald test.

Age: 57 y (24 y)*0.690.45–1.07.098
Presentation: recurrence (primary)2.371.46–3.86.001
Syndrome: NF-1 (no NF-1)1.090.62–1.93.759
Tumor site: head and neck (extremity)1.810.52–6.22.256
Tumor site: trunk (extremity)1.510.90–2.52 
Tumor size: 12 cm (4 cm)1.691.33–2.14<.001
Grade 3 (Grade 1 or 2)2.161.08–4.29.028
Margin status: positive (negative)1.310.79–2.18.290
Radiation therapy: yes (no)0.750.47–1.21.239
Chemotherapy: yes (no)1.490.88–2.53.141

DISCUSSION

In this series of >200 patients with localized MPNST at any site who underwent surgery with intent to eradicate disease at our institution during a 25-year period, the cumulative incidence of both local recurrence and distant metastasis at 10 years was approximately 30%. Disease-specific mortality at 10 years after definitive surgery was 43%, but the probability of achieving a cure was no greater than 40%. These results are similar to those reported in the main published series (Table 6) and reflect the known biologic aggressiveness of this rare disease.8–16 Therefore, we analyzed a number of factors that had a possible influence on the clinical outcomes of patients with these tumors.

Table 6. 5-Year and 10-Year Incidence Estimates in Major Published Series
ReferenceNo. of patientsYears of study5 years (%)10 years (%)
LRDMCSMLRDMCSM
  • LR indicates local recurrence; DM, distant metastasis; CSM, cause-specific mortality; MSKCC, Memorial Sloan Kettering Cancer Center, New York, NY; NR, not reported; MGH, Massachusetts General Hospital, Boston, MA; INT, Istituto Nazionale Tumori, Milan, Italy.

  • *

    Without neurofibromatosis type 1 (NF-1).

  • With NF-1.

  • Total.

Sordillo et al., 1981 (MSKCC)9165NR  47*  32*
2318
Ducatman et al., 1986 (Mayo Clinic)101201912–1983422834  22
Hruban et al., 1990 (MSKCC)11431945–1988406539  NR
Wanebo et al., 1992 (Bethesda, MD)12281960–1990NRNR43NRNRNR
Doorn et al., 1995 (Groningen, the Netherlands)14221977–1990NRNR35NRNR24
Wong et al., 1998 (Mayo Clinic)151341975–1993494952454034
Cashen et al, 2004. (MGH)16801972–1997NRNR81*NRNRNR
Current study (INT)2051976–200227.326.239.928.828.743.3

Presentation with either primary or recurrent disease, tumor size, and the site of origin were identified as the most important prognosticators for cause-specific survival in patients with MPNST. Patients who presented at our institute with a local recurrence had worse outcomes compared with patients who presented with primary disease. This negative impact of presentation was significant for all 3 of the endpoints investigated. We observed that patients who presented with a local recurrence were at an increased risk of developing another local recurrence, distant metastases, and eventually dying of their disease. This difference well may reflect a selected subset of patients and may suggest that local recurrences need to be followed even more carefully, because these patients have a 2.5-fold risk of mortality from the disease.

Tumor size was investigated as a continuous variable, and we confirmed that its negative impact increased continuously with increasing size: the larger the tumor, the worse the outcome. This negative impact was related both to local recurrence and distant metastases. Patients with larger tumors had reduced local control and distant control rates and, thus, had an increased risk of mortality. This observation is in keeping with some of the results from major series (Table 7).

Table 7. Significant Prognostic Factors in Major Published Series
ReferenceYearNo. of patientsPresentationSiteSizeGradeNF-1
  1. NF-1 indicates neurofibromatosis type 1; MSKCC, Memorial Sloan Kettering Cancer Center, New York, NY; +, with NF-1; NR, not reported; −, without NF-1; MGH, Massachusetts General Hospital, Boston, MA; INT, Istituto Nazionale Tumouri, Milan, Italy.

Sordillo et al., 1981 (MSKCC)91980165 +NRNR+
Ducatman et al., 1986 (Mayo Clinic)101986120 ++
Hruban et al., 1990 (MSKCC)11199043 ++
Wanebo et al., 1993 (Bethesda, MD)12199228 ++
Doorn et al., 1995 (Groningen, the Netherlands)14199522 NRNRNR
Wong et al., 1998 (Mayo Clinic)151998134 ++
Cashen et al., 2004 (MGH)16200480 +NR+
Current study (INT)2005205+++

With regard to location, patients who had tumors of the extremities had more favorable outcomes compared with patients who had tumors of the trunk and of the head and neck. This difference was related mainly to a difference in local control. Tumors of the trunk bore a 3-fold risk of recurring locally compared with tumors of the extremity. Therefore, patients who had tumors of the trunk were at 2.5-fold risk of dying of disease, mainly from locoregional failure, as also reported in many series of patients with trunk and retroperitoneal soft tissue sarcomas.21, 22 In the group of patients with tumors of the head and neck (only 9 patients), the small numbers made it difficult to draw any conclusions for this site. Nevertheless, those patients had the worst outcome, mainly related to the worst local control. This observation was likely to be true, because head and neck tumors share similarities with other trunk locations and have even closer proximity to vital structures, making local control more critical for survival. The importance of tumor site also was evident in other series (see Table 7). Two additional factors were identified that affected survival in our series, almost exclusively through a negative impact on locoregional outcome: margin status and radiotherapy.

Patients who had positive surgical margins bore a 2.4-fold risk of developing a local recurrence and a 1.8-fold risk of dying of disease. The same difference was not observed in patients who had metastases. Therefore, patients who had positive surgical margins had a greater risk of dying, mainly from locoregional disease. It is noteworthy that this effect was independent from tumor location and size and was not linked to the development of subsequent metastases, which may be true for patients with other types of soft tissue sarcoma.23 This observation was in keeping with several other studies, which reported the importance of surgical margin status in predicting local recurrence and possibly survival among patients with soft tissue sarcoma.24–28

Patients who received radiation therapy had approximately 50% of the risk of developing a local recurrence compared with patients who did not received radiation therapy, although the difference was not statistically significant, and they bore 50% of the risk of dying from their disease. No major difference in metastases among the irradiated and not irradiated groups were observed. It also is worth noting that the impact on survival was independent of margin status. Therefore, a recommendation to consider radiation therapy for all operated localized patients could be made, although it would be better to have a formal prospective study to clarify this particular aspect.

Tumor grade was not identified as a significant factor for survival in the multivariate analysis. This lack of significance well may reflect a different proportion of patients in the 3 different categories: most of our patient were classified with high-grade tumors (70%), and only 6% of patients had low-grade tumors. Nevertheless, we observed that patients who had high-grade tumors bore a 1.8-fold risk of dying from their disease compared with patients who had low-grade tumors. This difference could be explained mainly by a 2-fold risk of developing distant metastases in patients who had high-grade tumors compared with patients who had Grade 1 and 2, and this remained significant in the multivariate analysis. Wong et al.15 also observed that tumor grade had prognostic significance, although those authors did not use the FNCLCC system for grading assessment. It is noteworthy that their criteria for grading included cellularity, nuclear pleomorphism and anaplasia, mitotic rate, microvascular proliferation, and necrosis, most of which also are included in the French classification system. Hruban et al.11 reported that patients who had tumors with >20 mitoses per 10 high-power fields had a much worse survival rate compared with patients who had tumors with fewer mitoses.

Twenty-two percent of patients in our series had NF-1. Other studies in the literature reported that from 21% to 70% of their study populations were affected. No doubt patients with NF-1 are at increased risk of developing MPNST. In our study, there was no significant difference in the survival of patients with or without NF-1, even if there was a trend toward an increased risk of local recurrence and metastases in the patients with NF-1. Patients with NF-1 had larger tumors on average (10 cm) than the other patients (6 cm). The importance of NF-1 also was reported in other series (Table 7). Our current analysis suggests that the unfavorable outcome of patients with NF-1 is related mainly to the development of larger tumors rather than the presence of underlying biologic factors related to the disease. Patients without NF-1 seem to present at an earlier stage, whereas patients with NF-1 tend to present late, because they probably fail to recognize a malignancy early among the benign tumors they already have and keep developing through the years.

In the current study, we observed that patients with MPNST shared prognostic factors similar to those observed in patients with other soft tissue sarcomas. The results confirmed that these patients have among the worst clinical outcomes, as reported previously by some authors, who designed tools for predicting outcomes based on large case-series analyses.29–31 Presentation with either primary or recurrent disease, tumor size, and tumor location have major importance with regard to prognosis. Complete tumor resection with negative margins should be the objective of surgery. Adjuvant radiation therapy should be delivered to improve local control and also may be beneficial for survival. High-grade and large MPNSTs have particularly aggressive behavior and, thus, patients with these tumors deserve to be considered for new adjuvant medical treatments. To discover these tumors as early as possible, patients with NF-1 syndrome should be followed carefully because of the likelihood that they will develop MPNSTs.

Ancillary