Proposed adjustments to pathologic staging of epithelial malignant pleural mesothelioma based on analysis of 354 cases

Authors

  • William G. Richards PhD,

    Corresponding author
    1. Division of Thoracic Surgery, Brigham and Women's Hospital, Boston, Massachusetts
    2. International Mesothelioma Program, Brigham and Women's Hospital, Boston, Massachusetts
    • Brigham and Women's Hospital, Thoracic Surgery, 75 Francis Street, Boston, MA 02115
    Search for more papers by this author
    • Fax: (617) 264-6810

  • John J. Godleski MD,

    1. International Mesothelioma Program, Brigham and Women's Hospital, Boston, Massachusetts
    2. Department of Pathology, Brigham and Women's Hospital, Boston, Massachusetts
    Search for more papers by this author
  • Beow Y. Yeap ScD,

    1. International Mesothelioma Program, Brigham and Women's Hospital, Boston, Massachusetts
    2. Department of Medicine, Massachusetts General Hospital, Boston, Massachusetts
    Search for more papers by this author
  • Joseph M. Corson MD,

    1. International Mesothelioma Program, Brigham and Women's Hospital, Boston, Massachusetts
    2. Department of Pathology, Brigham and Women's Hospital, Boston, Massachusetts
    Search for more papers by this author
  • Lucian R. Chirieac MD,

    1. International Mesothelioma Program, Brigham and Women's Hospital, Boston, Massachusetts
    2. Department of Pathology, Brigham and Women's Hospital, Boston, Massachusetts
    Search for more papers by this author
  • Lambros Zellos MD, MPH,

    1. Division of Thoracic Surgery, Brigham and Women's Hospital, Boston, Massachusetts
    2. International Mesothelioma Program, Brigham and Women's Hospital, Boston, Massachusetts
    Search for more papers by this author
  • Aneil Mujoomdar MD,

    1. Division of Thoracic Surgery, Brigham and Women's Hospital, Boston, Massachusetts
    2. International Mesothelioma Program, Brigham and Women's Hospital, Boston, Massachusetts
    Search for more papers by this author
  • Michael T. Jaklitsch MD,

    1. Division of Thoracic Surgery, Brigham and Women's Hospital, Boston, Massachusetts
    2. International Mesothelioma Program, Brigham and Women's Hospital, Boston, Massachusetts
    Search for more papers by this author
  • Raphael Bueno MD,

    1. Division of Thoracic Surgery, Brigham and Women's Hospital, Boston, Massachusetts
    2. International Mesothelioma Program, Brigham and Women's Hospital, Boston, Massachusetts
    Search for more papers by this author
  • David J. Sugarbaker MD

    1. Division of Thoracic Surgery, Brigham and Women's Hospital, Boston, Massachusetts
    2. International Mesothelioma Program, Brigham and Women's Hospital, Boston, Massachusetts
    Search for more papers by this author

Errata

This article is corrected by:

  1. Errata: Erratum: Proposed adjustments to pathologic staging of epithelial malignant pleural mesothelioma based on analysis of 354 cases Volume 116, Issue 10, 2503, Article first published online: 11 March 2010

Abstract

BACKGROUND:

Several pathologic staging systems for malignant pleural mesothelioma (MPM) have been published, but none of them provide optimal survival stratification or stage distribution among patients treated with surgery. Interpretation of prior studies that correlate pathologic factors with outcome has been confounded by the inclusion of patients undergoing differing surgical procedures and with varied tumor histology.

METHODS:

We examined pathologic characteristics, previously included in published studies, and explored correlations with outcome among patients with epithelioid MPM who underwent extrapleural pneumonectomy (EPP) at Brigham and Women's Hospital (BWH). Comparisons of survival among patients with and without each tumor or lymph node feature guided adjustments to the American Joint Commission on Cancer (AJCC)/International Union Against Cancer (UICC) classification criteria. Proportional hazards modeling of TN combinations guided adjustments to stage groupings.

RESULTS:

Three hundred fifty-four patients were resectable by EPP and had complete pathologic data. Overall median survival was 18 months from surgery. By AJCC/UICC criteria, 233 (66%) patients were stage III, whereas by BWH criteria, 194 (55%) patients were stage III. T classification criteria were adjusted based on prevalence and relation to survival. N status correlated significantly with survival. Regrouping of TN combinations based on relative hazard and Kaplan-Meier survival analysis resulted in improved stage distribution (stage I-IV: 8%, 43%, 33%, 16%, respectively) and survival stratification (51, 26, 15, 8 months, respectively).

CONCLUSIONS:

Proposed adjustments to TNM staging criteria improved outcome stratification of patients with epithelial tumor histology who received surgical therapy by EPP and complete pathologic assessment. Determining relevance to other treatment or staging modalities will require verification in additional cohorts. Cancer 2010. © 2010 American Cancer Society.

Developing an accurate pathologic staging system to predict outcome for patients with diffuse malignant pleural mesothelioma (MPM) has been challenging because of the rarity of this malignancy and the heterogeneity of treatment approaches.

MPM is classified based on histologic criteria into epithelioid, sarcomatoid, biphasic, and desmoplastic subtypes.1 Epithelioid MPM is associated with a more favorable prognosis than the nonepithelial subtypes as documented in multiple studies.2-5 Essentially all prior outcome analyses in MPM have involved patient cohorts with various combinations of tumor subtypes, confounding attempts to evaluate more subtle prognostic factors relevant to staging.

Developing an accurate pathologic staging system requires a sufficient degree of surgical resection to permit assessment of critical indicators of malignant progression. However, the majority of published series have included patients with incomplete resection, staged retrospectively by relying on the surgeons' description of disease extent in situ.5, 6 Determining the prognostic significance of specific variables has been further complicated by differences in pathological assessment and reporting between and within institutions. For these reasons, the pathologic staging systems that have been proposed for MPM have not provided accurate stratification of patients in terms of survival.

Butchart7 proposed the first pathologic staging system for MPM in 1976, which was based on the application of a single therapy: extrapleural pneumonectomy (EPP). This pioneering effort was limited by high-operative mortality, poor overall survival, and small sample size. We proposed modifications to the Butchart staging system based on a larger series treated at Brigham and Women's Hospital (BWH) with EPP-based trimodality therapy.4, 8, 9 Involvement of surgical margins and extrapleural lymph nodes with tumor were identified as important prognostic factors.4 However, the Butchart and BWH staging systems are limited in that they do not conform to the TNM format ubiquitously used to stage other solid tumors. Furthermore, by defining stage IV exclusively in terms of distant metastasis, they do not adequately identify those surgically resectable patients with poor prognosis.

The International Mesothelioma Interest Group (IMIG) staging system,10 later adopted by the American Joint Commission on Cancer (AJCC) in the sixth edition of its Cancer Staging Manual11 and by the International Union Against Cancer (UICC),12 placed pathologic staging of MPM in an explicit TNM framework. Stage groupings were based on the extensive experience of TNM staging of lung cancer. However, the utility of the current TNM system to correlate stage with survival has not been documented, except to differentiate early (stages I and II combined) from advanced (stages III and IV combined) disease.5, 6, 13 The prognostic utility of the current AJCC/UICC system is further limited in that two-thirds of patients undergoing EPP with complete pathologic examination of the specimen are classified as stage III despite a broad range of survival duration.13

Our objective was to evaluate the prognostic significance of each of the individual pathological factors that comprise the classification criteria of the BWH and AJCC/UICC staging systems in a large cohort of patients with available outcome data. It is important to note that the pathologic workup of EPP specimens at BWH has followed a specific and highly detailed standardized protocol.14 The protocol has been in place consistently throughout the period of time used for this study and has included the systematic evaluation of lymph nodes and specific margins. To minimize confounding influences on prognosis, only tumors with epithelial histology were studied, and complete resection by EPP was required to enable comprehensive pathological evaluation of each specimen. We sought to determine whether data-driven adjustments to the AJCC/UICC TNM classification and stage grouping criteria would improve stage distribution and survival stratification.

MATERIALS AND METHODS

Data were obtained by retrospective medical record review and accumulated into the International Mesothelioma Program Patient Data Registry, with approval from the BWH Institutional Review Board (Protocol 2005-p-001520). Before outcome analysis, all pertinent pathologic data fields were subjected to a comprehensive audit, with consensus review and resolution of ambiguities in the clinical record by the Program pathologists (JC, JG, LC). In some cases, this process required retrieval and re-review of archival slides, with particular reference to the classification of epithelial histology according to World Health Organization criteria (<10% sarcomatoid component).1

The standard procedures followed at BWH for the pathological examination of EPP specimens include explicit guidelines for the pathological analysis of gross specimens, a sign out checklist, and a dictation template.14 Briefly, the specimen (comprising the lung, pleurae, diaphragm, and pericardium en bloc) is inflated, fixed, and coronally sectioned. A standard set of perpendicular margins is taken (anterior, posterior, medial, lateral, and apical pleural; anterior, posterior, medial, lateral, and inferior diaphragmatic). In addition, areas of gross tumor encroachment on diaphragm, pericardium, lung parenchyma, lobar fissures, chest wall, and mediastinal soft tissue or other structures are sampled for microscopic examination. Meticulous dissection of lymph nodes in the hilar region is undertaken.

This uniform, comprehensive analysis and reporting procedure enabled the unambiguous retrospective audit of key pathologic features in the Patient Data Registry required for the current analysis. The involvement of tumor at each margin or structure was scored present or absent. All features listed as T classification criteria in either the AJCC/UICC or the BWH staging system were included in the analysis (Table 1). For N classification, the numbers of nodes sampled and of nodes with microscopically documented tumor metastases were recorded for each nodal station, including levels 2-1115 and the internal mammary nodes. For the subset of patients who had cervical mediastinoscopy before EPP, lymph nodes that were biopsied during mediastinoscopy were included in the tallies.

Table 1. Observed Prevalence, Survival Analysis, and Proposed Adjustments to T Classification for Involvement of Anatomic Features by the Primary Tumor (N=354)
Extent of tumor invasionPrevalenceMedian SurvivalHRClassification
InvolvedUninvolvedCurrentProposed
  1. HR indicates hazard ratio; BWH, Brigham and Women's Hospital.4

<1 mm from any margin310 (88%)17.049.92.1BWH IT2
Lateral pleural margin130 (37%)17.119.11.1BWH II+T2
Posterior pleural margin123 (35%)18.817.81.0BWH II+T2
Fissures292 (82%)17.025.81.6T2T2
Pericardium (nontransmural)201 (57%)18.018.11.1T3T2
Endothoracic fascia127 (36%)16.519.61.1T3T2
Chest wall (solitary)117 (33%)16.518.81.1T3T2
Diaphragmatic margin74 (21%)13.719.11.3BWH II+T3
Apical pleura margin122 (34%)15.821.01.5BWH II+T3
Medial pleural margin147 (42%)16.819.21.3BWH II+T3
Anterior pleural margin131 (37%)16.219.81.3BWH II+T3
Lung parenchyma195 (55%)15.422.51.4T2T3
Diaphragm muscle137 (39%)14.621.01.4T2T3
Diaphragm204 (58%)17.021.01.3T2T3
Chest wall muscle39 (11%)13.418.81.2T3T3
Previous chest tube site34 (10%)12.219.21.6BWH IT4
Bronchial margin11 (3%)9.618.41.8BWH II+T4
Mediastinal fat35 (10%)12.319.21.5T3T4
Chest wall (diffuse)33 (9%)10.019.21.7T4T4
Through diaphragm (peritoneum)19 (5%)11.918.52.1T4T4
Internal surface of pericardium24 (7%)13.318.71.6T4T4

Survival duration was determined by subtracting the date of surgery from the date of death or last follow-up. Kaplan-Meier survival estimates and univariate hazard ratios were used to determine the prognostic significance of tumor involvement for each T classification feature and lymph node region.

The method used to assign individual T features to classification levels was based on a model that assumes a progressive accumulation of negative prognostic features as a growing tumor advances from one T classification level to the next. T1 tumors are characterized by the absence of negative prognostic features (ie, involvement of pleura only). T2 tumors are characterized by T2-level features only, T3 tumors by a combination of T2 and T3 features, and T4 tumors by T2, T3, and T4 features. The model, therefore, assumes a high prevalence of T2 features (eg, involvement of intralobar fissures is a common feature among T2, T3, and T4 tumors) with decreasing prevalence of features associated with T3 and T4 classification levels.

Each pathologic T feature examined was assigned to a classification level according to its prevalence and its hazard ratio as an indicator of poor prognosis within the cohort. No adjustment was made to the previously described T classification level for features that were observed at very low prevalence (<1%). Among features with hazard ratios ≥1.3, those observed at low prevalence were designated T4-level criteria, those at moderate prevalence were designated T3-level criteria, and those at high prevalence were designated T2-level criteria. Any feature not significantly related to survival (hazard ratio <1.3) was assigned to a T level 1 lower than indicated by its prevalence.

Regional lymph nodes were designated according to the AJCC/UICC 6th edition system.11, 12 Hilar and intraparenchymal lymph nodes were designated N1. Ipsilateral or midline mediastinal nodes and internal mammary nodes were N2. Contralateral mediastinal or hilar and any scalene or supraclavicular nodes were N3.

Proportional hazards modeling was used to assess the joint effects of T and N classification on overall survival. Dummy indicator variables were used to model each of the classification levels to obviate the assumption of a linear effect. Hazard ratios were derived for the possible TN combinations based on multiplying the model estimates associated with the respective T and N classification levels. The model was parameterized so that T1N0 served as the reference group with a hazard ratio of unity.

All P values are 2-sided. StatView version 4.5 software (Abacus Concepts, Piscataway, NJ) was used for all statistical calculations.

RESULTS

Four hundred seventy-three patients with epithelial MPM underwent surgical exploration for planned EPP between 1988 and 2005. Fifty-five patients were found to be unresectable, and another 53 were converted to pleurectomy when it was determined that macroscopic complete resection could not be achieved. Among the remaining 365 patients who underwent EPP, 3 patients lacked information on lymph node status (NX) and 8 had incomplete assessment of surgical margins (TX). Complete T and N classification data were available for 354 patients, who comprised the study cohort. For the cohort, 256 (72%) were male, the median age was 57 years, and 30-day or in-hospital mortality was 4.2% (n = 15). The median survival for this cohort of patients with epithelial tumors treated with EPP was 18 months from surgery. Fifty-three (15%) patients were alive (censored) with a median follow-up interval of 30 months from surgery (Fig. 1).

Figure 1.

Overall survival after surgery.

Application of BWH staging criteria4 (Fig. 2a) identified 158 patients (45%) as having stage I or II disease with favorable prognosis (40-month and 22-month median survival, respectively). One hundred ninety-four (55%) patients were classified as stage III. Survival was stratified by stage up to 3 years after surgery, but estimated survival functions for stages II and III merged after 40 months, and long-term survival associated with both stages was similar. Only 2 patients were classified as stage IV, which in this system requires demonstration of metastatic disease. In contrast, the AJCC/UICC criteria11, 12 classified 69 patients or 19% of the cohort as stage IV, whereas only 9 patients were classified as stage I, 43 patients were classified as stage II, and 233 (66%) patients were classified as stage III. Estimated survival for patients with stages II and III merged after 40 months (Fig. 2b).

Figure 2.

(A) Survival by the Brigham and Women's Hospital staging system.4 Two patients with stage IV disease (not shown) are deceased with survival durations of 1.9 and 2.1 months. (B) Survival by the American Joint Commission on Cancer/International Union Against Cancer staging system.11, 12

The prevalence of primary tumor involvement with specific anatomical structures and associated patient survival patterns are presented in Table 1. Seven AJCC/UICC T4 classification criteria (bone, mediastinal organs, contralateral pleura, spine, malignant pericardial effusion, myocardium, brachial plexus) were observed at less than 1% prevalence in the cohort and were, therefore, not analyzed. The remaining pathologic features sorted into clusters with low (3%-11%), moderate (21%-58%), and high (82%-88%) prevalence. Six low-prevalence, 7 moderate-prevalence, and 2 high-prevalence features were significantly associated with poor survival and were assigned to T4, T3, and T2 classification levels, respectively. One low-prevalence and 5 moderate-prevalence features were not significantly associated with survival (hazard ratio <1.3) and were, respectively, assigned to T3 and T2 levels.

The resulting data model suggested specific adjustments and additions to the AJCC/UICC T classification criteria. Involvement of lung parenchyma and diaphragm were moved from T2 to T3, and involvement of mediastinal fat was moved from T3 to T4. Three criteria, including tumor involvement of endothoracic fascia, chest wall at a single focus and into (but not through) pericardium, were not significantly related to survival and, thus, were moved from T3 to T2. Several pathologic features were added at appropriate T classification levels, and included tumor within 1mm of any margin (T2), involvement of specific surgical margins (T3, T4), and tumor within prior chest tube sites (T4). Together, these additions and adjustments yielded the proposed T classification criteria presented in Table 2.

Table 2. T Classification Criteria With Proposed Modifications
T1Tumor involves any of the ipsilateral pleural surfaces with negative resection margins (>1 mm)
T2Tumor present within 1 mm of any resection margin, involving endothoracic fascia or chest wall soft tissue (localized), penetrating into but not through pericardium, or involving intralobar fissures
T3Invasion of chest wall muscle, invasion into but not through diaphragm, invasion of lung parenchyma, or involvement of the apical, anterior, or medial pleural, or any diaphragmatic resection margin
T4Diffuse chest wall invasion, seeding of tumor in a prior chest tube site, transdiaphragmatic invasion, involvement at the bronchial resection margin, invasion of mediastinal organs or adipose tissue, direct extension to contralateral pleura, rib involvement, invasion of spine, extension through pericardium, malignant pericardial effusion, invasion of myocardium or brachial plexus.

Published AJCC/UICC N classification criteria correlated with survival in this cohort (Fig. 3a). However, the effect of nodal involvement on outcome could be further stratified by lymph node level within the current N2 classification (Table 3). Patients with involvement limited to inferior mediastinal and/or internal mammary nodes (here designated N2a) had significantly longer survival duration relative to those with involvement of lymph nodes in superior mediastinal stations (N2b; Fig. 3b).

Figure 3.

(A) Survival by the American Joint Commission on Cancer/International Union Against Cancer N classification.11, 12 One patient with N3 involvement (not shown) is deceased with survival duration of 0.2 month. (B) Survival by N classification with proposed N2 stratification: N2a = levels 5-9 and internal mammary nodes; N2b = levels 2-4.15

Table 3. Observed Prevalence, Survival Analysis, and Proposed Adjustments to N Classification for Tumor Involvement at Ipsilateral Lymph Node Levels15
Lymph Node LevelNodes SampledCases Involved/Median SurvivalHRPClassification
Median (Range)Sampled (%)InvolvedNot InvolvedCurrentProposed
  1. HR indicates hazard ratio; int. mamm., internal mammary lymph nodes.

Level 2-42 (1-17)24/83 (29%)6.223.53.7<.0001N2N2b
Level 5-9 + int. mamm.3 (1-27)196/210 (46%)11.923.52.0<.0001N2N2a
Level 10-147 (1-37)170/354 (48%)14.323.01.6<.0001N1N1

Proportional hazards modeling was used to estimate the hazard ratio associated with each of the TN combinations. A single patient with N3 disease was omitted from the model to avoid computational instability. The hazard ratios were used to rank the T-N combinations in order of ascending prognosis. The selection of cutpoints for stage grouping was guided by natural clustering of hazard ratios and median Kaplan-Meier survival duration (Fig. 4). This analysis led to recommended adjustments to stage grouping criteria (Table 4). Insufficient data were available to evaluate N3 and M1 disease, which remained as stage IV criteria.

Figure 4.

Hazard ratio and Kaplan-Meier median survival (n) for subsets of patients grouped by combined T and N classification using modified criteria. For cells with insufficient events for valid Kaplan-Meier analysis, patient status (A = alive, D = deceased) and survival duration in months are provided. Recommended stage groupings based on hazard ratio and survival are indicated by cell color: Red indicates stage I; orange, stage II; green, stage III; blue, stage IV.

Table 4. Proposed TNM Stage Grouping
StageTumorNodeMetastasis
I   
 T1-2N0M0
 T1N1-2aM0
II   
 T1N2bM0
 T2N1M0
 T3N0-1M0
III   
 T2-3N2aM0
 T4N0-1M0
IV   
 T2-3N2bM0
 T4N2M0
 T1-4N3M0
 T1-4N0-3M1

The proposed modifications to the TNM system resulted in significant stratification of survival by stage with improved stage distribution when applied to the current cohort (Fig. 5). Similar numbers of early and late stage patients were identified, with median survival durations of 51 months for 28 stage I patients, 26 months for 151 stage II patients, 15 months for 117 stage III patients, and 8 months for 58 stage IV patients.

Figure 5.

Survival by proposed modified TNM staging system.

DISCUSSION

The goal of establishing a staging system for MPM is to allow the physician to more accurately predict which patients will benefit from a particular therapeutic intervention. The system should permit the stratification of patients for accurate evaluation of new therapeutic strategies in terms of outcome endpoints and the selection of patients most likely to benefit from established therapies, particularly in the adjuvant setting. Pathologic staging systems have the benefit of directly documenting the extent of malignant progression. For many solid tumors, the disease process has been effectively modeled in terms of pathologic assessment of the primary lesion, lymphangitic spread and hematogenous dissemination, expressed in a TNM framework.

During the 1990s, the IMIG proposed modifications16 to the AJCC (fourth edition) TNM system,17 which incorporated prognostic factors that had been identified in several published series of MPM patients. As validation of the modified staging system, a survival difference has been demonstrated between stages I-II and stages III-IV in several studies,5, 6, 13 but not between individual stages. The patient cohorts examined in these studies were not ideal for validating a pathologic staging system. They included all MPM histologic subtypes and surgical procedures that varied from exploration to pleurectomy/decortication and EPP. For example, in the initial validation study,6 131 consecutive patients underwent exploratory thoracotomy, but only 101 patients went on to have resection and only 50 of those had EPP. Thus, 81 patients were, at best, staged surgicopathologically by the gross evaluation of tumor extent reported by the surgeon in the operative note. In addition, 36 patients had nonepithelial MPM. Imbalance of this potent prognostic factor could confound comparison of stage groups, particularly in smaller studies. The current analysis demonstrates that even within a large, histologically homogeneous cohort of patients with complete pathological evaluation, the current TNM criteria do not provide for stage-by-stage survival stratification. The potential requirement for future validation of and possible adjustments to TNM criteria as additional data would become available was well recognized and described in the IMIG and AJCC publications.6, 11, 16

The practical use of the T1a/T1b distinction for pathologic staging is unclear. Disease limited to the parietal pleura appears to be quite rare among surgical patients. Two patients in the current cohort fit the T1a classification and both experienced extended survival relative to 6 patients with T1b tumors. In the Memorial Sloan-Kettering study,6 18 of 131 tumors were classified as T1, but the numbers of T1a versus T1b were not reported. Similarly, because observation of nodal involvement with T1 tumors was exceedingly rare in the current analysis (0.56%), the Cox model's predicted hazard ratios for T1N1-2a classifications are theoretical, and the resulting assignment of these classifications to stage I cannot be assessed. It would seem intuitively appropriate to assign these patients to stage Ib or even stage II, although as a practical matter the issue is moot.

The BWH staging system4, 8 evolved as an extension of the Butchart system.7 Both systems were based exclusively on patients treated with EPP. The BWH system emphasized the prognostic importance of surgical margins and lymph node involvement. There is still considerable variability among institutions in the assessment and reporting of surgical margins and hilar/intraparenchymal lymph nodes. Accurate staging requires complete assessment of histological subtype, tumor extent, resection margins, and nodal metastases. The BWH protocol requires routine sampling of specific margins and meticulous dissection of hilar and peribronchial lymph nodes.14 Recommendations published by the College of American Pathologists18 and the Association of Directors of Anatomic and Surgical Pathology19 include taking perpendicular sections of margins where involvement is suspected by gross criteria and analysis of all “received” lymph nodes as required elements for reporting of EPP specimens. The present analysis re-emphasizes the prognostic implications of specific surgical margins and N1 status as justification for the incremental time and labor required for thorough assessment of specimens.

Although each of the 2 published pathological staging systems examined here stratified survival in the current cohort for the first 36 months, there was less consistent stratification of the 28% of patients who survived beyond 3 years. For example, these patients were assigned primarily to stage III and some to stage IV by AJCC/UICC criteria. In addition, each system was limited in that a majority of patients were classified as stage III. Brigham criteria identified more stage I-II patients with favorable prognosis, while AJCC/UICC criteria identified more stage IV patients. Data-driven adjustment of AJCC/UICC T and N classification and stage grouping criteria resulted in improved stage distribution and stratification of survival when the new criteria were reapplied to the patient cohort. Importantly, most patients surviving beyond 3 years are classified to stages I and II. These modifications incorporate surgical margins and emphasize nodal status, integrating the primary elements of BWH criteria4 into the previously published TNM framework.11, 12

The AJCC/UICC staging system grouped N1 and N2 disease together within stage III, because available data at the time the IMIG recommendations were published did not support a survival distinction among N classifications.10 The current study demonstrates that N2 disease represents a more advanced stage with poorer prognosis than does N1, an observation supported by prior work.4, 20 The further novel finding reported here, that the involvement of superior mediastinal nodal stations portends particularly poor prognosis after EPP, supports the recommendation that N2 classification might be usefully divided, particularly in the context of advanced T status. Flores et al20 reported N2 involvement in 110 of 348 (32%) cases, in agreement with the 106 of 354 (30%) observed in the present study. N2 sampling has generally been performed by the surgeon on the basis of the gross appearance of nodes suspicious for malignant involvement. By contrast, N1 involvement was observed in 170 of 354 (48%) cases in the current study with uniform complete resection and pathologic evaluation of N1 nodes, compared with only 87 of 348 (25%) reported in the Flores et al study, which included patients resected with pleurectomy/decortication for whom N1 evaluation is not possible.20 This observation highlights the potential for incomplete staging of patients for whom complete microscopic examination of disease extent is not possible owing to the retention of the lung and hilar structures within the patient, as occurs during pleurectomy/decortication.

More generally, the current analysis suggests that epithelial MPM progresses from the diaphragmatic pleura to the apex of the chest. Pathologists have previously observed that the bases are more extensively involved with disease than the apices.14 To the extent that the disease may depend on pleural fluid as a vehicle of intrapleural dissemination of inhaled asbestos fibers and/or shed tumor cells, this finding is consistent with a gravitational influence on the tumor's natural history. The current study documents a worsening prognosis as tumor involves lymph nodes and surgical margins at superior thoracic levels. These observations also support the utility of mediastinoscopy to assess superior N2 and N3 nodal involvement21 and careful radiographic assessment of the apical chest when considering EPP as primary therapy.

Staging systems must evolve to account for the interaction between tumor biology and the therapeutic intervention being applied. The adjustments to classification and stage grouping criteria suggested here improved survival stratification in this cohort of patients with epithelial tumor histology who received definitive surgical therapy by EPP. Application of other therapeutic strategies, including pleurectomy/decortication, will address the generic extent of the proposed adjustments. We suspect that one would be able to use this procedure with mediastinoscopy to establish stage with a level of confidence, although the use of such procedures, and the strong divergence of clinical outcomes among histologic subtypes, may require further adjustments or annotation of the staging of MPM. The prognostic significance of individual TNM factors may be affected as nonepithelial disease is considered and as new therapies are developed that interact with tumor biology.

Acknowledgements

The authors acknowledge Carl Alsup and Jordan Mueller for assistance in establishing the Patient Data Registry, Dr. John A. Mannick for critical review of the manuscript, and Ann S. Adams for editorial assistance.

CONFLICT OF INTEREST DISCLOSURES

Supported by the International Mesothelioma Program.

The authors made no further disclosures.

Ancillary