(Chest. 2004;125:116S-119S.)
© 2004
American College of Chest Physicians
Elevated Hepatocyte Growth Factor Level Correlates With Poor Outcome in Early-Stage and Late-Stage Adenocarcinoma of the Lung*
Jill M. Siegfried, PhD;
James D. Luketich, MD;
Laura P. Stabile, PhD;
Neil Christie, MD and
Stephanie R. Land, PhD
* From the Departments of Pharmacology (Drs. Siegfried and Stabile), Surgery (Drs. Luketich and Christie), and Biostatistics (Dr. Land), University of Pittsburgh Cancer Institute, Pittsburgh, PA.
Correspondence to: Jill M. Siegfried, PhD, The Hillman Cancer Center, UPCI Research Pavilion, Suite 2.18, Pittsburgh, PA 15213-1863; e-mail: siegfriedjm{at}upmc.edu
 |
Introduction
|
|---|
The 5-year survival rate for patients with all stages of lung cancer combined is only 15%. The survival rate is 48% for patients in whom disease is localized when detected, demonstrating that even when diagnosis occurs at an early stage of disease, relapse and death are common. Thus, identifying prognostic markers is critical for predicting patient survival in early disease and for determining proper therapeutic strategies. Hepatocyte growth factor (HGF) is a pleiotropic protein that induces cell proliferation and cell movement,12 and is a powerful angiogenic factor.3 It is primarily a paracrine factor that is produced by mesenchymal cells,4 although carcinoma cells may secrete HGF.56 The receptor for HGF, c-Met, is expressed by both epithelial and endothelial cells. The HGF/c-Met signaling pathway plays a significant role in the pathogenesis of many human cancers and is an attractive target for lung cancer therapy.7 The overexpression of HGF has been observed in either serum or tumor extracts of patients with lung cancer.89
We previously published a study8 on the association of elevated HGF with poor outcome in patients with non-small cell lung cancer who underwent lung resection. These data suggested that HGF is a negative prognostic indicator for lung cancer. We have now refined the previous study to include only patients in whom lung adenocarcinoma has been diagnosed, have enlarged it to include 15 additional adenocarcinoma patients, and have restricted it to patients who underwent curative resection, with complete mediastinal and hilar lymph node dissections for pathologic staging. We also incorporated updated outcome information for the cases included previously. For all patients, an HGF level above the median was significantly associated with shorter all-cause survival time (p = 0.016), shorter lung cancer survival time (p = 0.004), and shorter disease-free survival time (p = 0.001). Analyzing stage I patients alone, an HGF level above the median was significantly associated with shorter lung cancer survival time (p = 0.027) and shorter disease-free survival time (p = 0.007). In a multivariate analysis of lung cancer survival, the three variables that were most prognostic were HGF (dichotomous), age, and stage. HGF was also an independent prognostic variable.
|
Analysis of HGF in Tumor Tissues
|
|---|
Tumor tissue was collected sequentially from patients with institutional review board approval, as described previously,8 and was analyzed for HGF protein.8 Analysis was restricted to patients for whom follow-up information was available and who had documented pathology reports from a complete mediastinal and hilar lymph node resection. This report included the analysis of 45 adenocarcinomas meeting these criteria from our previous publication, with additional follow-up on the 30 patients still living at the time of the last analysis. The mean follow-up period for patients still living in the previous report was 29 months. Here, the mean follow-up time for patients still living was 61 months (median, 55 months; range, 16 to 112 months).
|
Clinical Data and Statistical Analysis
|
|---|
The diagnosis of adenocarcinoma primary to the lung (adenocarcinoma, 48 cases; adenosquamous carcinoma, 6 cases; bronchioloalveolar carcinoma, 5 cases) and the pathologic staging of lung tumors was confirmed by pathology reports. Recurrences were documented by physical examination, diagnostic procedures, and radiologic tests, and deaths were confirmed by tumor registry data. The following three survival end points were analyzed: (1) time from curative resection to death from any cause (ie, all-cause survival); (2) time from curative resection to death from lung cancer (ie, lung cancer survival); and (3) time from curative resection to lung cancer recurrence (ie, disease-free survival). Patients were compared with the log-rank test and Kaplan-Meier survival curves.10 Analysis of the prognostic importance of HGF level, controlling for other variables, was performed with Cox proportional hazards regression.11 Candidate prognostic variables that were highly associated, such as stage and nodal status, were compared using the Schwarz criteria of the univariate models.12 The Schwarz criteria were used to compare these alternative groupings of categoric variables. Analyses were performed separately for all-cause survival, lung cancer survival, and disease-free survival. The relationship between HGF and other variables was examined with t tests (for continuous variables) and 
2 tests (for categoric variables). Survival trees were estimated by recursive partitioning performed with statistical software (rpart freeware; Mayo Foundation for Medical Education and Research, Rochester, MN),13 which is based on the classification and regression trees methodology of Breiman et al.14
|
HGF Levels and Clinical Profile of Patients
|
|---|
The range of HGF protein (0.7 to 198 ng HGF per 40 µg protein) showed a nonnormal distribution (not shown). The median HGF level was 22.4 ng per 40 µg protein, and the mean HGF level was 34 ng per 40 µg protein. Because the HGF level showed a nonnormal distribution, the median was a more appropriate choice for the survival analysis. We also performed a tree analysis to determine the optimum value for the dichotomization of HGF. In various analyses for survival, the optimal value for splitting HGF was the median. HGF level was not significantly associated with gender, stage, type of therapy, nodal status, T status, degree of differentiation of the tumor, or smoking history. The estimated follow-up time by the Kaplan-Meier reverse censoring method (censoring at death) was not significantly different between low and high HGF levels.
|
Survival Analysis
|
|---|
Stage was a significant factor in all survival measurements, indicating that, within our cohort, the expected effect of advanced disease on survival was observed. The survival of patients with stage IA, IB, II, and IIIA disease combined were statistically different (by log-rank test) for disease-free survival (p = 0.0007), for lung cancer survival (p = 0.0182), and for all-cause survival (p = 0.0177). In examining HGF status, 18 of 25 patients who were alive with no evidence of disease (NED) fell into the low-HGF group. Two patients who died of other causes and had NED at death also fell into the low-HGF group. Of those patients who had died of lung cancer, 17 of 22 fell into the high-HGF group. There was a significant association between HGF category and vital status (ie, alive vs dead from any cause during the follow-up period, p = 0.02 [Fisher exact test]), between HGF category and death from lung cancer (ie, alive or dead NED vs dead of lung cancer during follow-up period, p = 0.003 [Fisher exact test]), and HGF category and survival without recurrence during follow-up period (alive NED vs other, p = 0.004 [Fisher exact test]).
|
Survival by HGF Within Disease Stages
|
|---|
Table 1
displays the events and the p values for the log-rank test between high and low HGF strata (
22.4 vs > 22.4, respectively expressed as ng/40 µg protein) for all patients combined, for those with stage IA and IB disease combined, and for those with stage IB disease. All-cause survival, lung cancer survival, and disease-free survival end points were significantly different between the high-HGF and the low-HGF groups for all patients combined. This is illustrated in the Kaplan-Meier survival plots for all patients combined (Fig 1
). For all stage I patients, lung cancer survival and disease-free survival showed a significant difference between the high-HGF and low-HGF groups. All-cause survival was suggestive (p = 0.055) but not significant. Figure 2 illustrates the Kaplan-Meier survival plots for patients with stage I disease.
View larger version (19K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1. Kaplan-Meier plots comparing patients with HGF levels above and below the median (22.4) with respect to all-cause survival (top left, A), lung cancer survival (top right, B), and disease-free survival (bottom, C).
|
|
View larger version (19K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2. Kaplan-Meier plots comparing patients with stage I disease with HGF levels above and below the median (22.4) with respect to all-cause survival (top left, A), lung cancer survival (top right, B), and disease-free survival (bottom, C).
|
|
|
Multivariate Survival Analyses
|
|---|
Nodal status, T-status, smoking history, age, disease stage, prior therapy, gender, and HGF status (dichotomous) were analyzed. For all-cause survival, the variables that were significant and yielded the best fit with the model in univariate models were as follows: nodal status; age; and HGF status. For lung cancer survival, the variables that were significant and yielded the best fit with the model in univariate models were as follows: stage; age; and HGF status. For disease-free survival, nodal status and HGF were the selected variables. There was no significant association between HGF levels and the other selected prognostic variables as follows: age (t = 1.22; p = 0.28); nodal status (2 = 0.30; p = 0.58); and stage (2 = 0.62; p = 0.73). HGF was a significant factor in the multivariate analysis of lung cancer survival (p = 0.035) and disease-free survival (p = 0.003). Relative risks indicated that the effect of HGF was clinically as well as statistically significant (for all-cause survival, 2.2; for lung cancer survival, 3.0; and for disease-free survival, 3.3).
|
Conclusions
|
|---|
The data presented here confirm the association of elevated HGF with poor survival in patients with lung cancer, specifically in those with cases of adenocarcinoma of the lung. HGF level was not associated with stage, nodal status, T status, or any other clinical parameter examined. Thus, HGF is a strong independent negative prognostic predictor at all stages of disease for adenocarcinoma, as well as in stage IA and IB disease. Stage I patients with elevated HGF levels in the primary resected lung adenocarcinoma had a significantly higher risk of recurrence and death from lung cancer compared to those with low HGF levels. These data confirm those from our previous report that HGF could be used to identify patients with stage I disease who are at higher risk for recurrence and who could be candidates for adjuvant therapy.
|
Footnotes
|
|---|
Abbreviations: HGF = hepatocyte growth factor; NED = no evidence of disease
This research was supported by grants RO1 CA 79882 and P50 CA090440 (Specialized Program of Research Excellence in Lung Cancer) awarded to Dr. Siegfried from the National Cancer Institute. Dr. Stabile was supported by a fellowship from the American Lung Association.
|
References
|
|---|
- Nakamura, T, Nawa, K, Ichihara, A (1984) Partial purification and characterization of hepatocyte growth factor from serum of hepatectomized rats. Biochem Biophys Res Commun 122,1450-1459[CrossRef][ISI][Medline]
- Morimoto, A, Okamura, K, Hamanaka, R, et al M. Hepatocyte growth factor modulates migration and proliferation of human microvascular endothelial cells in culture. Biochem Biophys Res Commun 1991;179,1042-1049[CrossRef][ISI][Medline]
- Bussolino, F, Di Renzo, MF, Ziche, M, et al Hepatocyte growth factor is a potent angiogenic factor which stimulates endothelial cell motility and growth. J Cell Biol 1992;119,629-641[Abstract/Free Full Text]
- Rosen, E, Meromsky, L, Setter, E, et al Smooth muscle-derived factor stimulates mobility of human tumor cells. Invasion Metastasis 1990;10,49-64[ISI][Medline]
- Tsao, MS, Zhu, H, Giaid, A, et al Hepatocyte growth factor/scatter factor is an autocrine factor for human normal bronchial epithelial and lung carcinoma cells. Cell Growth Differ 1993;4,571-579[Abstract]
- Tokunou, M, Niki, T, Eguchi, K, et al c-MET expression in myofibroblasts: role in autocrine activation and prognostic significance in lung adenocarcinoma. Am J Pathol 2001;158,1451-1463[Abstract/Free Full Text]
- Singh-Kaw, P, Zarnegar, R, Siegfried, JM Stimulatory effects of hepatocyte growth factor on normal and neoplastic human bronchial epithelial cells. Am J Physiol 1995;268,L1012-L1020
- Siegfried, JM, Weissfeld, LA, Singh-Kaw, P, et al Association of immunoreactive hepatocyte growth factor with poor survival in resectable non-small cell lung cancer. Cancer Res 1997;57,433-439[Abstract/Free Full Text]
- Takigawa, N, Sewaga, Y, Maeda, Y, et al Serum hepatocyte growth factor/scatter factor levels in small cell lung cancer patients. Lung Cancer 1997;17,211-218[CrossRef][ISI][Medline]
- Kaplan, EL, Meier, P Nonparametric estimation for incomplete observations. J Am Stat Assoc 1958;53,457-481[CrossRef][ISI]
- Cox, DR Regression models and life tables. J R Stat Soc Ser B 1972;34,187-220
- Schwarz, G Estimating the dimension of a model. Ann Stat 1978;6,461-464
- Therneau TM, Atkinson EJ. Technical report series no. 61: An introduction to recursive partitioning using the RPART routines. Rochester, MN: Department of Health Science Research, Mayo Clinic, 1997. Available at: http://www.mayo.edu/hsr/techrpt.html. Accessed April 14, 2004
- Breiman, L, Friedman, JH, Olshen, RA, et al Classification and regression trees 1984 Wadsworth Publishing Co Inc. Belmont, CA:
TOP
Introduction
Analysis of HGF in...
