Volume 133, Issue 6 , Pages 1419-1427.e4, June 2007
Fluorine-18 fluorodeoxyglucose positron emission tomographic maximal standardized uptake value predicts survival independent of clinical but not pathologic TNM staging of resected non–small cell lung cancer
Article Outline
Objectives
Positron emission tomographic maximal standardized uptake value has been shown to predict survival after resection of non–small cell lung cancer. The relative prognostic benefit of maximal standardized uptake value with respect to other clinical/pathologic variables has not been defined.
Methods
We reviewed patients who had positron emission tomographic imaging and an R0 resection for non–small cell lung cancer between January 1, 2000, and December 31, 2004, without induction or adjuvant therapy. The associations between overall survival, histology, pathologic TNM stage, pathologic tumor diameter, and standardized uptake value were tested.
Results
Four hundred eighty-seven patients met the study criteria. Median follow-up was 25.8 months. By using the median values for tumor size (2.5 cm) and standardized uptake value (5.3), standardized uptake value was an independent predictor of survival (P = .03), adjusting for tumor size (P = .02) and histology (P < .01). The optimal standardized uptake value for stratification was identified as 4.4, and this value was identified as an independent predictor of survival (P = .03) after adjusting for clinical TNM stage. Standardized uptake value was not an independent predictor of survival (P = .09), adjusting for pathologic TNM stage (stage IA vs IB vs stage II–IV, P < .01).
Conclusions
Standardized uptake value does not add to the prognostic significance of pathologic TNM stage. Standardized uptake value was an independent prognostic factor from clinical TNM stage.
CTSNet classification: 10
Abbreviations and Acronyms: CT, computed tomography, 18F-FDG, fluorine-18 fluorodeoxyglucose, NSCLC, non–small cell lung cancer, pTNM, pathologic TNM, SUV, standardized uptake value, SUVMAX, maximal standardized uptake value
The standardized uptake value (SUV) for fluorine-18 fluorodeoxyglucose (18F-FDG) as measured by using positron emission tomography (PET) has been shown to correlate with several measures of tumor behavior, such as lesion doubling time1 and Ki-67 staining,2 suggesting that SUV might be a predictor of patient prognosis. Previously, to determine whether 18F-FDG uptake in a malignancy correlated with prognosis, we performed a retrospective review of patients with histologically proved non–small cell lung cancer (NSCLC) or carcinoid cancer (pathologic T1-4N0-2M0) who had undergone R0 resections after PET imaging and without either neoadjuvant or adjuvant therapy.3 We found that stratification of patients by the median SUVMAX (which was 9) predicted survival; the 2-year survival for patients with an SUVMAX of greater than 9 was 68%, and that for patients with an SUVMAX of less than 9 was 96% (P < .01, log–rank test). However, an insufficient number of patients was available to allow an analysis of the relationship of SUVMAX to pathologic TMN staging. Other reports have attempted to determine whether SUVMAX is an independent predictor from pathologic TNM (pTNM) staging of survival, but none have contained sufficient patients to allow a definitive answer.4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 Results in the published studies have been mixed, and possible reasons for the conflicting results have been recently extensively analyzed in a review by Pillot and coauthors.16 To analyze whether PET SUV was a predictor of survival independent of pTNM staging, we reviewed our experience with a much larger cohort of patients treated since our original report.
Earn CME credits at http://cme.ctsnetjournals.org
Materials and Methods
We performed a retrospective review of patients who had PET imaging and an R0 resection for NSCLC without induction or adjuvant therapy at Memorial Sloan-Kettering Cancer Center during the period from January 1, 2000, through December 31, 2004. The primary goal of this study was to determine whether SUVMAX of the primary site of disease was an independent predictor of survival from clinical and pTNM staging.
This review was performed after approval had been obtained from the Memorial Sloan-Kettering Cancer Center Institutional Review Board and in accord with an assurance filed with and approved by the Department of Health and Human Services.
Patient data were obtained from a prospectively maintained database in the Memorial Sloan-Kettering Cancer Center Thoracic Service, in which patient staging information is entered on a weekly basis under attending surgeon supervision, and survival data were updated at regular intervals by research study assistants. The primary end point was overall survival, which was calculated from the date of surgical intervention to the date of death or last contact with the patient.
Histologic characterization of the tumors was obtained from the operative pathologic report. Patients undergoing resections of what were believed by the treating physicians to be synchronous or metachronous (defined for the purpose of this study as within 5 years of prior lung cancer) primary NSCLCs were not included. Tumors were grouped into 3 categories: adenocarcinoma, squamous cell carcinoma, and “other.” Patients who had no nodes sampled at the time of resection were treated as having pathologic N0 disease. Adenosquamous carcinomas were included in the “other” category. If recorded, SUVMAX of the primary tumors was obtained from the radiology report, and if not recorded, it was calculated by a Nuclear Medicine physician (TA) from PET images. For statistical analyses, clinical and pTNM stages were compressed to 3 groups: stage IA, stage IB, and stage II to IV.
Associations between overall survival and tumor histology, clinical TNM stage, pTNM stage, and pathologic maximal tumor diameter were tested by using the log–rank test, and the associations between survival and SUVMAX of the primary site of disease were evaluated with Cox regression. In determining the interrelation between SUV and variables that can be determined preoperatively, we used pathologically measured tumor size determined from pathology reports as a surrogate for imaging-based estimates of tumor size and pathologic histology as a surrogate for histology determined from a fine- or core-needle biopsy. Estimation of the optimal values for stratification was performed by using the maximal χ2 method. Multivariate modeling to identify independent prognostic factors was performed by using Cox regression. Poor, intermediate, and good risk groups were identified by merging subgroups with similar subgroups through backward elimination following the Cox model. Thirty-three patients in the current cohort had been included in the previously reported analysis3; analyses were performed with these patients excluded, and the results did not differ significantly from those found in the material presented below (data not shown).
Results
Four hundred eighty-seven patients met the study criteria. Patient demographics are recorded in TABLE 1A, TABLE 1B, TABLE 1C. There was a predominance of male subjects (67%) and of adenocarcinoma (69%). The majority of patients (87.5%) underwent anatomic resections. Pathologic stage was IA in 249 patients, IB in 132 patients, and IIA through IV in 106 patients. With a median follow-up of 25.8 months, 17 (3%) patients have been lost to follow-up.
TABLE 1A. Patient demographics
| Sex | |
 Male subjects | 230 |
| Female subjects | 257 |
| Age (y), median (range) | 69(27–87) |
| Extent of resection | |
| Bilobectomy | 18 |
| Lobectomy | 347 |
| Pneumonectomy | 19 |
| Segmentectomy | 41 |
| Wedge | 62 |
| Histology | |
| Adenocarcinoma | 337 |
| Squamous | 104 |
| Other | 46 |
| Pathologic tumor size (mm), median (range) | 25(4–140) |
| SUVMAX primary tumor | 5.3(0.6–36.3) |
| Follow-up (mo), median (range) | 25.8(1–66) |
| Status at last follow-up | |
| Alive with disease | 38 |
| Dead of disease | 69 |
| Dead of other causes | 20 |
| No evidence of disease | 343 |
| Lost to follow-up | 17 |
TABLE 1B. Clinical TNM stage
| Clinical stage | |
| T | |
| T1 | 326 |
| T2 | 153 |
| T3 | 5 |
| T4 | 3 |
| N | |
| N0 | 404 |
| N1 | 54 |
| N2 | 26 |
| N3 | 2 |
| M | |
| M0 | 476 |
| M1 | 4 |
| Clinical TNM | |
| IA | 284 |
| IB | 105 |
| IIA | 20 |
| IIB | 34 |
| IIIA | 26 |
| IIIB | 6 |
| IV | 4 |
| Not available | 8 |
TABLE 1C. Pathologic TNM stage
| Postresection stage | |
| T | |
| T1 | 283 |
| T2 | 173 |
| T3 | 13 |
| T4 | 18 |
| N | |
| N0 | 395 |
| N1 | 55 |
| N2 | 27 |
| N3 | 0 |
| None sampled | 10 |
| M | |
| M0 | 479 |
| M1 | 8 |
| Pathologic TNM | |
| IA | 249 |
| IB | 131 |
| IIA | 21 |
| IIB | 35 |
| IIIA | 27 |
| IIIB | 16 |
| IV | 8 |
By using univariate analysis, tumor size (Figure 1) and SUV were determined to be significant predictors of survival (P < .02 and .03, respectively). Survival after resection stratified by histology (Figure 2) demonstrated significant differences between adenocarcinoma and squamous carcinoma (P = .05), adenocarcinoma and other histologies (P < .01), and squamous carcinoma and other histologies (P = .05).
Figure 1.
Survival after resection stratified by optimal value for maximal tumor diameter.
Figure 2.
Survival after resection stratified by histology (P < .01 for adenocarcinoma [Adeno] vs other, P = .05 for adenocarcinoma vs squamous, P = .05 for squamous vs other).
The median SUV for adenocarcinoma was significantly different from that for the squamous carcinomas (P < .01) and that for the other histologies (P < .01). The median SUV values for squamous carcinoma and for other histologies were not different (P = .69, Figure E1).
Figure E1.
Histograms of distribution of standardized uptake value (SUV) within each histologic type. The median SUV for adenocarcinoma (Adeno) was significantly different from the squamous and other histologies (P < .01). The median values for squamous carcinoma and histologies categorized as “other” were not different (P = .69).
We first performed an analysis of significant prognostic variables using the median values for SUV and for pathologic tumor size to avoid inadvertent bias. In an analysis using the median values for pathologically measured tumor size (2.5 cm) and tumor SUVMAX (5.3), SUVMAX was an independent predictor of survival (P = .03) after adjusting for tumor size (P = .02) and histology (P < .01). After demonstrating that the median value for SUV was a significant independent predictor of survival, the optimal cut-off point for stratification was calculated as a tumor size of 3.3 cm (P < .03) and an SUV of 4.3 (P < .01, Figure 3). This optimal value for SUV was an independent predictor of survival (P = .03) after adjusting for clinical TNM stage (as stage IA vs IB vs II-IV, P < .01). There was an interaction between SUV and clinical TNM stage in that SUV acted most strongly in the stratification of clinical TNM stage IB (Figure E2). Survival stratified by a combination of the optimal values for SUV and size demonstrated that patients with a size of greater than 2.5 cm and an SUV of greater than 4.3 had a significantly worse survival when compared with all other patients (Figure 4).
Figure 3.
Survival after resection stratified by optimal value for maximal standardized uptake value of the primary tumor.
Figure E2.
Survival curves for patients with clinical stage IA (top left), IB (top right), and II to IV (bottom left) disease stratified by the optimal value for standardized uptake value (SUV; 4.3) demonstrating interaction between SUV and clinical TNM stage.
Figure 4.
Survival stratified by a combination of the optimal values for standardized uptake value (SUV) and tumor size, demonstrating that patients with a tumor size of greater than 2.5 cm and an SUV of greater than 4.3 had significantly worse survival when compared with all other patients (P < .01).
The SUV in patients with pTNM stage IA disease was significantly lower than that in patients with stage IB disease (P < .01) and patients with stage II to IV disease (P < .01). The SUV distribution between stage IB disease and stage II to IV disease did not differ (P = .39, Figure E3). Survival after resection stratified by pTNM stage (Figure E4) demonstrated significant differences between pTNM stages IA, IB, and II to IV (P < .01). Neither the median value for SUV nor the calculated optimal SUV were independent predictors of survival (P = .09 for both) after adjusting for pTNM stage (stages IA vs IB vs II-IV, P < .01; Figure E5, Figure E6, Figure E7).
Figure E3.
Histograms of the distribution of standardized uptake value (SUV) within each stage group. SUV in stage IA disease is significantly lower than that in stage IB disease (P < .01) and stage II to IV disease (P < .01). SUV distribution in stage IB and stage II to IV disease was not different (P = .39).
Figure E4.
Survival after resection stratified by pathologic TNM stage (IA vs IB vs II-IV).
Figure E5.
Pathologic stage IA disease stratified by optimal standardized uptake value.
Figure E6.
Pathologic stage IB disease stratified by optimal standardized uptake value.
Figure E7.
Pathologic stage II to IV disease stratified by optimal standardized uptake value.
Given that SUV was an independent prognostic variable from clinical TNM stage, we attempted to determine whether there was a combination of prognostic variables available preoperatively, including SUV, that could approximate the prognostic information provided by the postoperative variables of pTNM staging and pathologic histology.
Combining histology and pTNM stage, we identified the following 3 statistically significant postoperative prognostic categories (Figure 5): good—adenocarcinoma, pTNM stage IA; poor—large cell/sarcomatoid, pTNM stages II to IV; intermediate, all other patients.
Figure 5.
Survival stratified by a combination of histology and pathologic TNM stage.
Combining SUVMAX, pathologic tumor size, and histology, we identified the following statistically significant preoperative prognostic categories (Figure 6): good—adenocarcinoma, SUV of less than 4.4 and size of less than 2.5 cm; poor—large cell/sarcomatoid histology, SUV of greater than 4.4 and size of greater than 2.5; intermediate—all others.
Figure 6.
Survival stratified by a combination of histology, standardized uptake value, and tumor size.
After complete pathologic staging, 394 (81%) of 487 patients remained in the same prognostic group that they had been assigned to on the basis of clinical staging. The 19% (93/487) of patients who change prognostic groups after resection do so primarily by moving from a good to an intermediate prognosis (Table 2). The 3-year survival for patients in the preoperative and postoperative good category was 86% and 87%, respectively; 72% and 67%, respectively, for the intermediate group; and 45% and 45%, respectively, for the poor group.
TABLE 2. Preoperative versus postoperative risk group stratification
| Preoperative (rows) vs postoperative (columns) | I | II | III |
|---|---|---|---|
| I | 190 | 72 | 0 |
| II | 12 | 177 | 6 |
| III | 0 | 3 | 27 |
Discussion
To determine whether PET SUV was a clinically relevant prognostic parameter, we examined a uniform population of patients who all had NSCLC of the lung resected with curative intent after PET imaging at one institution and who did not receive either induction or adjuvant therapy. Even though the number of patients available for analysis was more than twice the number (225) of surgically treated patients in the prior largest series,17 to provide the highest quality statistical analyses, we compared SUV as a prognostic variable against 3 prognostic groupings of TNM stages. After patients had been stratified into groups as having pTNM stage IA disease (small node-negative tumors), stage IB disease (large node-negative tumors), and stage II to IV disease (82/105 or 78% of whom had lymph node involvement), further stratification by means of PET SUV did not further define prognosis.
However, further stratification by means of PET SUV after clinical TNM staging significantly improved the prognosis. This suggests that there might be a role for PET SUV in defining patient prognosis preoperatively to help in determining which patients should be considered for induction therapy or for definitive chemoradiotherapy rather than surgical intervention. When we stratified patients on the basis of tumor maximum diameter and histology (both of which are variables that can be approximated by preoperative imaging and needle aspiration, respectively), the addition of PET SUV further improved the definition of prognosis. It is worth noting that because we used pathologic tumor size as a surrogate for radiographically determined size and final pathology as a surrogate for histology determined from a fine-needle aspiration biopsy, it is likely that when radiographic size and histology by fine-needle aspiration are compared with PET that the relative prognostic benefit of determining SUV will improve. Radiographic estimates of the size of tumors can vary depending on whether the adjacent lung is inflated at the time of sectioning. It is possible that image-based estimates of tumor size might vary, and therefore the utility of combining image-based tumor size with SUV to define prognosis should be tested formally.
In a model of how a preoperative stratification might work, we construct good-, intermediate-, and poor-prognosis groups based on the postoperative prognostic variables of pTNM stage and pathologic histology. Statistical analysis suggested that combinations of potentially available preoperative variables (SUV, tumor size, and histology) allowed construction of good-, intermediate-, and poor-prognosis groups that closely correlated with the postoperative categories in that only 19% of patients moved from one to another group preoperatively to postoperatively. This compares very favorably with the current relationship between clinical and pathologic staging, in which approximately 50% of patients will shift stage. For example, Roberts and colleagues18 found that T status as determined by means of computed tomographic (CT) imaging was concordant with pathologic T stage in only 56% of patients, with CT overstaging 20% and understaging 24%. Similarly, Cerfolio and coauthors17 found that after PET/CT and CT imaging, 52% of patients clinically staged as N2 positive were actually N2 negative, and 14% of patients clinically staged as N2 negative were actually pathologically N2 positive.
Validation of the utility of PET SUV as a prognostic variable and the relative benefit of SUV in comparison with imaging measurements and tumor characteristics, such as gene expression, has been undertaken in an attempt to better guide patient therapy at the time of diagnosis rather than after resection.
References
- . Lung tumor growth correlates with glucose metabolism measured by fluorine-18 fluorodeoxyglucose positron emission tomography. Ann Thorac Surg. 1995;60:1348–1352
- Lung cancer proliferation correlates with [F-18] fluorodeoxyglucose uptake by positron emission tomography. Clin Cancer Res. 2000;6:3837–3844
- Preoperative F-18 fluorodeoxyglucose-positron emission tomography maximal standardized uptake value predicts survival after lung cancer resection. J Clin Oncol. 2004;22:3255–3260
- . The prognostic significance of fluorodeoxyglucose positron emission tomography imaging for patients with nonsmall cell lung carcinoma. Cancer. 1998;83:918–924
- . The maximum standardized uptake values on positron emission tomography of a non-small cell lung cancer predict stage, recurrence, and survival. J Thorac Cardiovasc Surg. 2005;130:151–159
- . 18F]Fluorodeoxyglucose positron emission tomography and its prognostic value in lung cancer. Eur J Cardiothorac Surg. 2000;18:425–428
- 18F-FDG uptake as a biologic prognostic factor for recurrence in patients with surgically resection non-small cell lung cancer. J Nucl Med. 2002;43:39–45
- Determination of the prognostic value of [18F]fluorodeoxyglucose uptake by using positron emission tomography in patients with non-small cell lung cancer. Nucl Med Commun. 2002;23:865–870
- . Positron-emission tomography as a prognostic tool for early-stage lung cancer. Am J Surg. 2006;191:433–436
- Positron emission tomographic scanning in the diagnosis and staging of non-small cell lung cancer 2 cm in size or less. J Thorac Cardiovasc Surg. 2005;130:1611–1615
- [18F]fluorodeoxyglucose uptake by positron emission tomography predicts outcome of non-small-cell lung cancer. J Clin Oncol. 2005;23:1136–1143
- Prognostic importance of the standardized uptake value on 18F-fluoro-2-deoxy-glucose-positron emission tomography scan in non-small-cell lung cancer: An analysis of 125 cases. J Clin Oncol. 1999;17:3201–3206
- Does the FDG uptake of primary non-small cell lung cancer predict prognosis?: A work in progress. Clin Positron Imaging. 1999;2:111–118
- Relationship between non-small cell lung cancer fluorodeoxyglucose uptake at positron emission tomography and surgical stage with relevance to patient prognosis. Clin Cancer Res. 2004;10:4709–4716
- Standardised FDG uptake: a prognostic factor for inoperable non-small cell lung cancer. Eur J Cancer. 2005;41:1533–1541
- . Prognostic value of fluorodeoxyglucose positron emission tomography in non-small cell lung cancer (A review). J Thorac Oncol. 2006;1:152–159
- . Improving the inaccuracies of clinical staging of patients with NSLCC: a prospective trial. Ann Thorac Surg. 2005;80:1207–1213
- Prospective comparison of radiologic, thoracoscopic, and pathologic staging in patients with early non-small cell lung cancer. Ann Thorac Surg. 1999;68:1154–1158
Robert J. Downey
PII: S0022-5223(07)00235-8
doi:10.1016/j.jtcvs.2007.01.041
© 2007 The American Association for Thoracic Surgery. Published by Elsevier Inc. All rights reserved.
Volume 133, Issue 6 , Pages 1419-1427.e4, June 2007