Prognostic Modeling in Early Stage Lung Cancer: An Evolving Process from Histopathology to Genomics

https://doi.org/10.1016/j.thorsurg.2007.03.014Get rights and content

The goal is to validate a molecular-based tumor model that identifies patients at low-risk for cancer recurrence and who will not benefit from adjuvant chemotherapy. The remaining patients will be randomized to observation (present standard of care) or adjuvant chemotherapy to determine efficacy of adjuvant in this population. Investigators have focused on the identification of markers that may predict poor prognosis as a way to “enrich” the population by separating those likely to have early recurrence and cancer death from those not needing additional treatment after resection. The initial projects refined predictive models of cancer recurrence after resection for patients with early stage non–small cell lung cancer.

Section snippets

Histopathology and prognosis

A multitude of histopathologic factors were described in the literature from small, retrospective series that had an association with survival. This list was confusing; the author's laboratory undertook an investigation of all of these factors in a single, large cohort of NSCLC patients. Pathologic stage I was chosen to eliminate any interference from the powerful factors of distant and nodal metastases. No patients had adjuvant therapy after complete R0 resection. All tumors were recut and

Protein expression and immunohistochemistry

Although observational data based on clinical and histopathologic description aided in dividing patients into risk group, much variation exists in interpretation of these qualitative variables. Improvement in molecular techniques has allowed refinements in predictive models. Investigations in the author's laboratory focused on protein expression measured with immunohistochemistry. Multiple algorithms could be created based on relative protein expression profiles using up to 15 factors with

Building a novel genomic predictive model

A cohort of stage I patients was constructed with equal numbers of squamous cell carcinoma and adenocarcinoma, the histologies that constitute most NSCLC. Uniquely, equal numbers of patients with extremes of outcomes (good-risk tumors [those patients who lived >5 years after resection] and poor-risk tumors [those patients who died of metastatic cancer <2.5 years from resection]) were used to build the genomic model of tumor aggressiveness. The author's group has previously described methods to

Internal validation

The predictive accuracy of the metagene tree was assessed with various methods: internal validation (leave-one-out cross-validation and random test–training sets) and external validation (three separate non-Duke cohorts). Leave-one-out cross-validation is repeatedly performed; one sample is removed at each reanalysis and the recurrence probability is predicted for that one case. The entire model-building process is repeated for each prediction and evaluates the reproducibility of the approach.

External validation

Use of a prognostic model to assess risk of recurrence and to choose the appropriate use of adjuvant chemotherapy requires demonstration that the model is robust when applied to independent heterogeneous populations of patients and conditions of sample acquisition. Other investigators have developed predictive models for NSCLC, but little data are available that validate the models on external samples [21], [22], [23], [24], [25], [26]. The author's group evaluated the ability of the model to

Future directions

The goal of the investigations over the last decade has been to refine prognosis as an opportunity to select patients for a prospective phase III adjuvant chemotherapy clinical trial. CALGB 30506, “A Randomized Phase III Trial to Evaluate the Lung Metagene Score to Direct Adjuvant Therapy in Stage 1 NSCLC Patients,” is an intergroup-supported concept for a stage I NSCLC trial that uses the LMS to select appropriate patients for adjuvant therapy. The LMS is most accurate and specific for

References (27)

  • T. Winton et al.

    Vinorelbine plus cisplatin vs. observation in resected non-small cell lung cancer

    N Engl J Med

    (2005)
  • G.M. Strauss et al.

    Adjuvant chemotherapy in stage IB non-small cell lung cancer (NSCLC): update of Cancer and Leukemia Group B (CALGB) protocol 9633

    Proceedings of the American Society of Clinical Oncology

    (2006)
  • D.H. Harpole et al.

    A prognostic model of recurrence and death in stage I non-small cell lung cancer utilizing presentation, histopathology, and oncoprotein expression

    Cancer Res

    (1995)

    • Cited by (9)

    View all citing articles on Scopus
    View full text
    Copyright © 2007 Published by Elsevier Inc.