- Email: [email protected]
Repeat FDG-PET After Neoadjuvant Therapy is a Predictor of Pathologic Response in Patients With Non-Small Cell Lung Cancer
Robert J. Cerfolio, MD, Ayesha S. Bryant, MSPH, Thomas S. Winokur, MD, Buddhiwardhan Ohja, MD, MPH, and Alfred A. Bartolucci, PhD Department of Surgery, University of Alabama at Birmingham and Division of Cardiothoracic Surgery, Department of Surgery, Birmingham Veterans Administration Hospital; and the Departments of Epidemiology, Clinical Pathology, Nuclear Medicine, and Biostatistics, UAB School of Public Health, Birmingham, Alabama
Background. Repeat positron emission tomography (PET) with 18F-fluorodeoxyglucose (FDG) and chest computed tomography (CT) are used to assess the effectiveness of chemoradiotherapy in patients with non-small cell lung cancer (NSCLC); however, the change in the standardized uptake values (SUV) has not been correlated with the pathologic change of the primary tumor. Methods. This is a retrospective cohort study of a prospective database of 56 patients who had NSCLC, FDG-PET, and chest CT scans both before and after neoadjuvant therapy, followed by complete resection of their cancer. Maximum SUVs (maxSUV) and tumor size were measured, and the percentage of change was compared with the percentage of nonviable tumor cells. The primary objective was to measure the degree of correlation between these values. Results. The change in the maxSUV has a near linear
relationship to the percent of nonviable tumor cells in the resected tumors. FDG-PET’s maxSUV is better correlated to pathology than the change in size on CT scan (r2 ⴝ 0.75, r2 ⴝ 0.03, p < 0.001). When the maxSUV decreased by 80% or more, a complete pathologic response could be predicted with a sensitivity of 90%, specificity of 100%, and accuracy of 96%. Conclusions. The change in maxSUV on FDG-PET scan after neoadjuvant therapy holds a near linear relationship with pathologic response. It is a more accurate predictor than the change of size on CT scan. When the maxSUV decreases by 80% or more it is likely that the patient is a complete responder irrespective of cell type, neoadjuvant treatment, or the final absolute maxSUV. These findings may help guide treatment strategies. (Ann Thorac Surg 2004;78:1903–9) © 2004 by The Society of Thoracic Surgeons
L
that undergo resection have a significant survival advantage [11]. Thus, being able to identify responders leads to improved patient selection for surgery and may also help guide further treatment for patients with stage IV disease. Chest computed tomography (CT) and positron emission tomography (PET) using 18F-fluorodeoxyglucose (FDG) are most often used to assess the response of NSCLC to treatment. However little data have shown that the clinical response rate detected on repeat FDGPET scan actually correlates with the pathologic response of the primary tumor [12]. Therefore, we evaluated the effectiveness of repeat FDG-PET and repeat CT scan as predictors of this response in a group of patients who met strict entry criteria.
ung cancer is the leading cause of cancer deaths worldwide. More Americans die from lung cancer than the next three most common solid organ cancers (breast, prostate, and colon) combined [1– 4]. Because of these dismal numbers, more patients are undergoing neoadjuvant therapy before resection, adjuvant therapy after resection, and continued palliative chemotherapy. Several reports of prospective studies have shown increased survival favoring neoadjuvant therapy for patients with stage Ib, IIa, and N2 (stage IIIa) disease [5–9]. Patients with N2 disease who have had neoadjuvant therapy are usually only considered for resections if the N2 disease resolves. Similarly, further chemotherapy or changes in chemotherapy are often made for those patients with stage IV disease based on response. Improved survival has been shown in patients who have a response, even with stage IV disease [10]. Similarly, patients who are complete responders
Accepted for publication June 2, 2004. Address reprint requests to Dr Cerfolio, Division of Cardiothoracic Surgery, University of Alabama at Birmingham, 1900 University Blvd, THT 712, Birmingham, AL 35294; e-mail: [email protected].
© 2004 by The Society of Thoracic Surgeons Published by Elsevier Inc
Material and Methods Patients Over a 24-month period (January 2002 until December 2003), one general thoracic surgeon (RJC) performed 1,963 operations at the University of Alabama at Birmingham (UAB). Ninety-four patients had neoadjuvant therapy before resection, but only 56 met the strict 0003-4975/04/$30.00 doi:10.1016/j.athoracsur.2004.06.102
GENERAL THORACIC
Repeat FDG-PET After Neoadjuvant Therapy is a Predictor of Pathologic Response in Patients With Non-Small Cell Lung Cancer
GENERAL THORACIC
1904
CERFOLIO ET AL REPEAT FDG-PET AFTER NEOADJUVANT THERAPY
Ann Thorac Surg 2004;78:1903–9
Table 1. Entry Criteria
Table 2. Patient Characteristics
Patients were required to have the following: 1. Biopsy-proven NSCLC 2. An initial CT scan with 5 mm contiguous columnated slices with intravenous contrast prior to the start of neoadjuvant therapy. 3. An initial FDG-PET scan (on a dedicated FDG-PET scanner or an integrated PET-CT scanner) with the maxSUV measured prior to the start of neoadjuvant therapy. 4. The initial FDG-PET and chest CT scans had to be performed within 1 month prior to the start of the neoadjuvant therapy 5. Patients received neoadjuvant therapy (either using chemotherapy alone or combined chemoradiotherapy). 6. A repeat FDG-PET scan using the same PET center within 1 month of the completion of the neoadjuvant therapy using an identical dose of FDG with similar scanning techniques and with the maxSUV’s reported. 7. A repeat chest CT scan of the chest with 5 mm columnated cuts with intravenous contrast within 1 month of the completion of the neoadjuvant therapy. 8. A posterior-lateral thoracotomy with complete resection of the primary tumor along with complete thoracic lymphadenectomy within 5 weeks of the completion of the neo-adjuvant therapy.
Characteristics
CT ⫽ computed tomography; FDG-PET ⫽ 18F-fluorodeoxyglucose positron emission tomography; maxSUV ⫽ maximum standardized uptake values; NSCLC ⫽ non-small cell lung cancer.
entry criteria described in Table 1. The most common reason for exclusion of the other 38 patients was the lack of a dedicated FDG-PET scan or the lack of a maximum standardized uptake value (maxSUV) reported before the start of their neoadjuvant therapy. The remaining 56 patients comprised the cohort for this study. UAB’s Institutional Review Board approved this review of our prospective database. Patients who received neoadjuvant chemotherapy under the guise of the Southern Western Oncology Group S9900 trial received an additional separate consent. Patients were excluded if they had a history of type I diabetes mellitus, had a FDG-PET preformed on a nondedicated camera, did not have the maxSUV reported, had a delay greater than 6 weeks from the completion of their neoadjuvant treatment to resection, had an incomplete resection, were medically unfit, or refused surgery.
Imaging The PET scans performed at UAB were conducted on an integrated PET-CT scanner (GE Discovery LS PET-CT Scanner, Milwaukee, WI). Patients were asked to fast for 4 hours and then subsequently received 555 MBq (15 mCi) of FDG intravenously followed by PET after 1 hour. The maxSUV was determined by drawing regions of interest (ROI) on the attenuation-corrected FDG-PET images around the primary tumor. It was then calculated by the software contained within the PET-CT scanner by the formula: [13]
N ⫽ 56
Median age (range) Gender Male Female Neoadjuvant treatment type Chemoradiotherapy Chemotherapy only Median duration between last neoadjuvant treatment and surgery (range) Tumor histology Adenocarcinoma Squamous cell Other Preoperative cancer stage Ib–T2N0 IIa/IIb–T1, T2N1 IIb–T3N0 IIIa–T1, T2, T3N2 IIIa–T3N1 Stage IV–T2N0M1 (brain met) Type of operation Lobectomy Segmentectomy Pneumonectomy
63 (42–76) 31 (55%) 25 (45%) 23 (41%) 33 (59%) 29.8 days (20–40 days) 28 (50%) 22 (39%) 6 (11%) 11 (20%) 9 (16%) 8 (14%) 23 (41%) 4 (7%) 1 (2%) 43 (77%) 8 (14%) 5 (9%)
C (Ci ⁄ mL) MaxSUV ⫽
ID (Ci) w (kg)
where C ⫽ activity at a pixel within the tissue defined by an ROI and ID ⫽ injected dose per kg of the patient’s body weight (w). The maxSUV within the selected ROI’s was used. The percentage of change of the maxSUV was calculated by the formula: maxSUV initial ⫺ maxSUV final max SUV initial Patients that had FDG-PET scans performed at outside hospitals were included in this trial if the FDG-PET was performed on a dedicated PET camera with bismuth germanate crystals. All patients were carefully staged and biopsy specimens were taken from all suspicious N2, N3, or M1 areas (maxSUV ⬎ 2.5). Patients with N2 disease had chemoradiotherapy, and those that were N2 negative had chemotherapy alone. Biopsies were performed on all suspicious M1 metastatic lesions unless cancer was suspected in the bone or brain where magnetic resonance imaging (MRI) was considered to be the gold standard. Patients with suspected M1 disease in the liver, adrenal, or contralateral lung underwent definitive biopsy to prove or disprove M1 cancer. The volume of the tumor on chest CT was calculated using the equation:
CERFOLIO ET AL REPEAT FDG-PET AFTER NEOADJUVANT THERAPY
1905
Fig 1. Correlation between the decrease in maximum standardized uptake value (maxSUV) and the percentage of nonviable tumor in the resected specimens. The regression line equation: y ⫽ 0.9501 ⫻ ⫹5.230; r2 ⫽ 0.75, p ⬍ 0.001. Values where more than one data point was observed have been indicated by a solid black box and the number of observations are indicated in parenthesis. The open boxes represent one data point. Dashed lines indicate the 95% confidence interval for the regression line (solid line).
4 Volume ⫽ abc 3 with a, b, and c representing the height, diameter, and width of the mass (cm), respectively. This volume was calculated on the initial chest CT and then on the repeat chest CT after the neoadjuvant therapy was completed. The percentage of change in size was calculated by the formula: volume initial 共cm3) ⫺ volume final 共cm3) volume initial 共cm3)
Surgery Thoracotomy with complete thoracic lymphadenectomy was performed in all patients in this study. Patients who initially had stage IIIa disease, secondary to N2 disease, in general were offered pulmonary resection only if the initial nodal station involved with cancer was negative for cancer on repeat biopsy after the neoadjuvant therapy was completed. Repeat transesophageal ultrasound with fine needle aspirate was used to reassess lymph node stations 7, 8, and 9, and sometimes 5. Repeat videoassisted thorascopy was used to perform a repeat biopsy of stations 5 and 6. Repeat biopsies were performed on right-sided stations 2 and 4 and left-sided 4 by using open thoracotomy and frozen section analysis. Repeat mediastinoscopy was not performed. Segmentectomy, lobectomy, or pneumonectomy was performed; and wedge resection was not. The bronchial stump was buttressed with an intercostal muscle flap in all patients. As part of the entry criteria for this study, all patients had complete resection with negative margins and complete thoracic lymphadenectomy.
Pathologic Analysis Multiple hematoxylin and eosin stained sections of each tumor were reviewed by a pathologist and then rereviewed by another pathologist (TSW) who carefully calculated the percentage of nonviable tumor after recutting the entire tumor and scrutinizing multiple cross-
sections. The percent of nonviable tumor was defined as the combined percentage of scar and necrosis. Scar was defined as fibrous tissue intimately admixed with residual tumor. Patients with mixed tumors were labeled as squamous cell or adenocarcinoma based on the predominant cellular type seen. Complete pathologic response was defined as 1% or less of viable tumor cells detected on pathologic review of the entire resected specimens. The pathologists were blinded to all clinical, radiologic, and surgical findings.
Statistical Analysis The primary outcome evaluated was the degree of correlation between the percentage of change in maxSUV and the percentage of nonviable tumor in the resected specimens. We also evaluated the degree of correlation between the percentages of decrease of the volume of the mass on repeat CT scan and of nonviable tumor in the resected specimen. In addition, three secondary outcomes were also evaluated. The first was to assess the ability of FDG-PET and CT scan to predict complete pathologic response. The second was to compare the degree of correlation based on the type of neoadjuvant therapy. The third outcome assessed was to compare the degree of correlation based on the two main tissue types of cancer seen in this study, squamous cell and adenocarcinoma. Spearman’s rank-correlation coefficient was used to determine the relationship and establish a linear equation for estimating the percentage of nonviability of tumor according to the percentage of change in the maxSUV on FDGPET and also the change in volume of the mass on CT scan. Partial correlation coefficients were used to investigate relationships while adjusting for covariates. Multiple linear regression step-wise analysis was used to adjust for risk factors and identify any variables that were independently associated with a decrease in maxSUV (age, gender, percentage of nonviability of tumor, dose of radiation, cell type, and type of neoadjuvant therapy). The highest combined sensitivity and specificity values generated from the receiver operating characteristic (ROC) curves that predicted
GENERAL THORACIC
Ann Thorac Surg 2004;78:1903–9
GENERAL THORACIC
1906
CERFOLIO ET AL REPEAT FDG-PET AFTER NEOADJUVANT THERAPY
Ann Thorac Surg 2004;78:1903–9
Table 3. Actual Data Used to Compute Sensitivity, Specificity and Accuracy Values for Complete Responders for Decrease in Maximum Standardized Uptake Values and in Mass Volume on Computed Tomography FDG-PET maxSUV decrease
True Positive
False Positive
False Negative
True Negative
19 18 17 12
2 1 0 0
0 1 2 7
35 36 37 37
ⱖ60% ⱖ70% ⱖ80% ⱖ90% Decrease in volume on CT
Primary Outcomes
True Positive
False Positive
False Negative
True Negative
12 9 9 8
17 15 14 12
7 10 10 11
20 22 23 25
ⱖ60% ⱖ70% ⱖ80% ⱖ90%
CT ⫽ computed tomography; positron emission tomography; uptake values.
5 other patients without N2 disease received chemotherapy alone at outside institutions. Three of these patients had carboplatin and paclitaxel. Twenty-three patients with N2 disease underwent combined chemoradiotherapy, 18 of whom had carboplatin and paclitaxel. The dose of radiotherapy used was 45 to 51 Gy in 8 patients and 60 to 68 Gy in 15 patients. The median dose of preoperative radiation was 60 Gy (range 45 to 68 Gy). There was one operative death (1.8%) and no bronchopleural fistulas.
FDG-PET ⫽ 18F-fluorodeoxyglucose maxSUV ⫽ maximum standardized
complete pathologic responders were identified [14]. The ROC curves for FDG-PET and CT were compared according to the method of DeLong and associates [15]. The test for proportions and the binomial approximation test were used to compare the percent accuracy values between FDG-PET and CT scan. A two-sided p value of 0.05 or less was considered to indicate statistical significance. SAS version 9.0 (SAS Institute, Cary, NC) was used to conduct this analysis.
Results Patient Characteristics There were 56 patients (31 men) with a median age of 63 (range 42 to 76). Patient characteristics are shown in Table 2. Thirty-three patients had chemotherapy alone, and 23 patients with stage Ib, IIa, or non-N2 IIIa were enrolled in the Southern Western Oncology Group S9900 trial. All received carboplatin and paclitaxel. In addition,
Figure 1 demonstrates the percentage of change in the maxSUV plotted against the percentage of nonviable tumor for all patients. A near linear relationship is demonstrated (r2 ⫽ 0.75, p ⬍ 0.001). The correlation was stronger after adjusting for cell type, radiation dose, and type of neoadjuvant therapy (r2 ⫽ 0.81, p ⬍ 0.001). Age, gender, surgery type, and duration between the last dose of neoadjuvant therapy and resection were not significantly associated with the percentage of decrease in maxSUV. No correlation was evident between the change in tumor size on repeat CT scan and the percentage of nonviable tumor (r2 ⫽ 0.03, p ⬎ 0.05).
Secondary Outcomes/Complete Responders Nineteen patients were complete responders. Seven of the 33 patients (21%) who had chemotherapy achieved a complete response rate, whereas 12 of the 23 patients (52%) who had combined chemoradiotherapy had a complete pathologic response rate. The sensitivity and specificity of various cut points in the percentage of change of maxSUV and change in volume on CT scan were determined and are shown in Tables 3 and 4. We found that FDG-PET is superior to CT scan at all cut points for all variables (p ⬍ 0.05 for all). In addition, these variables are maximized when a cut point of 80% or greater is selected for the change in the maxSUV. Figure 2 shows the ROC curves that plot the sensitivity and specificity of the complete responders, and it again shows the superiority of FDG-PET (p ⬍ 0.001) over CT scan. In addition, we found no difference in the accuracy of the percentage of change of maxSUV for patients with N2 disease when compared with those without.
Table 4. The Sensitivity, Specificity and Accuracy of Predicting Complete Pathologic Response by the Percentage of Decrease in Maximum Standardized Uptake Values on Repeat FDG-PET and the Percentage of Decrease in the Mass Size of Tumor on CT Scana Sensitivity
ⱖ60% ⱖ70% ⱖ80% ⱖ90% a
Specificity
Accuracy
PPV
NPV
FDG-PET
CT
FDG-PET
CT
FDG-PET
CT
FDG-PET
CT
FDG-PET
CT
100% 95% 90% 63%
63% 47% 47% 42%
95% 97% 100% 100%
54% 59% 63% 68%
96% 96% 96% 88%
57% 56% 57% 59%
90% 94% 100% 100%
41% 38% 27% 40%
100% 97% 95% 84%
74% 69% 70% 69%
All p values are ⬍0.05.
FDG-PET ⫽ 18F-fluorodeoxyglucose positron emission tomography;
NPV ⫽ negative predictive value;
PPV ⫽ positive predictive value.
Fig 2. Receiver operating characteristic curves of sensitivity (solid lines) plotted against 100 minus specificity for 18F-fluorodeoxyglucose positron emission tomography (FDG-PET) and computed tomography (CT) for complete responders. (Area under curve ⫽ 0.935 for FDG-PET and 0.532 for CT, p ⬍ 0.001).
Type of Neoadjuvant Therapy FDG-PET was better able to predict the pathologic response in patients who had chemotherapy alone than in those that had chemoradiotherapy (r2 ⫽ 0.76 and 0.39, respectively; p ⬍ 0.001). Additionally, we observed a difference when we compared low-dose (ⱕ 51 Gy) radiation with high-dose (ⱖ 60 Gy) radiation (r2 ⫽ 0.90 and 0.66, respectively; p ⬍ 0.001). Repeat FDG-PET scan was a better predictor of pathologic response when lower doses of radiation were used.
Cell Type We found that FDG-PET was better able to predict the pathologic response in patients who had squamous cell cancer than in those that had adenocarcinoma (r2 ⫽ 0.83 and 0.58, respectively; p ⬍ 0.001). Interestingly, we also found that those with squamous cell cancer had a better response to their neoadjuvant therapy. The median decrease in maxSUV was 80% in patients with squamous cell cancer, and the median nonviable tumor found on pathologic examination was 88%. In contrast, the median decrease in maxSUV in patients with adenocarcinoma was 60%, and the median nonviable tumor found on pathologic examination for adenocarcinoma was 69%.
Comment Many studies have demonstrated the problem of falsepositive and false-negative results with FDG-PET scans in patients with NSCLC and the problems of restaging [16 –24]. One recurring theme is that FDG-PET scans do
CERFOLIO ET AL REPEAT FDG-PET AFTER NEOADJUVANT THERAPY
1907
not supplant the need for tissue biopsy specimens. We, along with others, have shown that a repeat FDG-PET scan may accurately predict the status of lymph nodes after neoadjuvant therapy [16, 24]; however, few reports have shown that repeat FDG-PET is an accurate predictor of the pathologic response of the primary tumor [12, 24, 25]. Nevertheless, it is commonly ordered, and the results are frequently used to guide, change, or even deny subsequent chemoradiotherapy or surgery, or both. This trial featured stringent entry criteria on a consecutive series of patients and showed that the change in maxSUV on a repeat FDG-PET scan is an accurate predictor of the pathologic response. The strength of this study, compared with other reports, is the entry criteria applied. For example, Ryu and colleagues in 2002 [24] retrospectively studied 26 patients and found that the change in SUVs of the primary tumor was useful for monitoring the therapeutic effect of neoadjuvant chemoradiotherapy in patients with NSCLC. Akhurst and colleagues in 2002 [25] reported a retrospective series on 56 patients and found that FDGPET after induction therapy accurately detected residual viable primary tumor. However, those reports did not require a baseline FDG-PET to show the percentage of change in maxSUV, a dedicated FDG-PET scan, or a separate tumor type analysis, as was done in our study. A limitation of our study is the short follow-up period and thus the lack of survival data. Yet others have shown that a significant response rate does equate to improved survival [10, 11]. Further studies with long-term analysis are required. The ability to noninvasively identify complete responders to neoadjuvant therapy could have important implications for surgical selection. Pisters and colleagues showed in 1993 [11] that the 5-year survival of patients who underwent resection and who had a complete pathologic response was 54% compared with 15% in those without a complete pathologic response. Nineteen (35%) patients in the study were complete responders, a similar incidence reported by Ryu and colleagues [24]. We found that when the maxSUV decreases by more than 80%, a complete responder can be predicted with 96% accuracy. In contrast, Port and colleagues [12] reported in 2004 that repeat PET did not predict the response to neoadjuvant therapy. However, in that report the authors arbitrarily defined a major PET response as a “reduction in the SUV of 50% or more.” If they had chosen a percentage of change of 80% or greater in maxSUV instead of 50% for their data they would have found that repeat PET correctly predicted 4 out of the 5 complete responders in their series. Importantly, we have demonstrated that the maxSUV does not need to fall to 0 to imply a complete responder, but rather it needs to have an 80% or greater reduction. In those patients who had preoperative chemotherapy, 7 patients were complete responders and in 5, the repeat maxSUV fell to 0. However, in the 12 complete responders who received combined chemoradiotherapy, only 6 had a maxSUV of 0 on the repeat FDG-PET. Thus, a patient who is a complete responder may have a maxSUV on the repeat FDG-PET greater than 0, especially if
GENERAL THORACIC
Ann Thorac Surg 2004;78:1903–9
GENERAL THORACIC
1908
CERFOLIO ET AL REPEAT FDG-PET AFTER NEOADJUVANT THERAPY
the patient had preoperative radiation. The percentage of drop is the best predictor of complete pathologic response and not the absolute value of the maxSUV on the repeat FDG-PET scan. One potential concern is that the SUV measured on one PET scanner may be different on another. The term standardized uptake value used to be an oxymoron because it was anything but standardized. However, most all dedicated PET scanners now use software packages that automatically calculate the maxSUV and take into account many of the various factors that can affect its value. Moreover, protocols for PET scanning are become more standardized, and centers are starting to use similar techniques of scanning, dosages of FDG, time intervals from injection to scanning, and baseline glucose, among others. These steps will make SUVs standardized across centers and continents. To protect for this potential problem in this study, we required that the patients received both FDG-PET scans at the same center using similar techniques. Finally, we chose to look at maxSUV instead of mean SUV because it eliminates many of the subjective measurements inherent to the latter. Interestingly we found that FDG-PET was less accurate for determining the pathologic response after neoadjuvant therapy when patients had radiation in addition to chemotherapy as opposed to chemotherapy alone. Other studies have demonstrated that radiation can interfere with the interpretation of FDG-PET scans [26]. We also found that the FDG-PET after neoadjuvant therapy was a more accurate predictor of response in patients who had squamous cell cancers than those who had adenocarcinoma, but the reasons for this are unclear. Our report shows that the percentage of change in the maxSUV on FDG-PET scan after neoadjuvant treatment is an accurate predictor of the actual pathologic response of the primary tumor in patients with NSCLC and that it also can identify complete responders. It is more accurate than CT scan. This information may help guide subsequent treatment strategies for medical oncologists, radiation oncologists, and surgeons. Although, this study and our conclusions must obviously be confined to patients with NSCLC, some of its findings may be applicable to patients with other types of solid organ cancers, especially ones that feature squamous cell or adenocarcinoma. Further prospective trials with long-term follow-up are needed.
References 1. Weir HK, Thun MJ, Hankey BF, et al. Annual report to the nation on the status of cancer, 1975–2000, featuring the uses of surveillance data for cancer prevention. J Nat Cancer Ins 2003;95(17):1276 –99. 2. Jemal A, Murray T, Samuels A, et al. Cancer statistics, 2003. CA Cancer J Clin 2003;53:5–26. 3. Parkin DM, Pisani P, Ferlay J. Estimates of the worldwide incidence of eighteen major cancers in 1985. Int J Cancer 1993;54:594 – 606. 4. Spira A, Ettinger DS. Multidisciplinary management of lung cancer. N Engl J Med 2004;3550:379 –92. 5. Depierre A, Milleron B, Moro-Sibilot, D, et al. Preoperative chemotherapy followed by surgery compared with primary
Ann Thorac Surg 2004;78:1903–9
6.
7.
8.
9.
10.
11.
12.
13. 14. 15.
16.
17. 18.
19. 20. 21.
22.
23.
surgery in resectable stage 1 (except T1N0), II and IIIa non-small cell lung cancer. J Clin Oncol 2002;20:247–53. Roth JA, Fossella F, Komaki R, et al. A randomized trial comparing perioperative chemotherapy and surgery with surgery alone in resectable stage IIIA non-small-cell lung cancer. J Natl Cancer Inst 1994;86(9):673– 80. Roth JA, Atkinson EN, Fossella F, et al. Long-term follow-up of patients enrolled in a randomized trial comparing perioperative chemotherapy and surgery with surgery alone in resectable stage IIIA non-small-cell lung cancer. Lung Cancer 1998;21(1):1– 6. Rosell R, Gomez-Codina J, Camps C, et al. Preresectional chemotherapy in stage IIIA non-small-cell lung cancer: a 7-year assessment of a randomized controlled trial. Lung Cancer 1999;26(1):7–14. Rosell R, Gomez-Codina J, Camps C, et al. A randomized trial comparing preoperative chemotherapy plus surgery with surgery alone in patients with non-small-cell lung cancer. N Engl J Med 1994;330:153– 8. Weber WA, Petersen V, Schmidt B, et al. Positron emission tomography in non-small-cell lung cancer: prediction of response to chemotherapy by quantitative assessment of glucose use. J Clin Oncol 2003;21(14):2651–7. Pisters KM, Ginsberg RJ, Giroux DJ, et al. Induction chemotherapy before surgery for early-stage lung cancer: a novel approach. Bimodality Lung Oncology Team. J Thorac Cardiovasc Surg 2000;119(3):429 –39. Port JL, Kent MS, Korst RJ, Keresztes R, Levin MA, Altorki NK. Positron emission tomography scanning poorly predicts response to preoperative chemotherapy in non-small cell lung cancer. Ann Thorac Surg 2004;77(1):254 –9. Nabi HA, Zubeldia JM. Clinical applications of F18-FDG in oncology. J Nucl Med Technol 2002;30:3–9. Grunkemeier GL, Jin R. Receiver operating characteristic curve analysis of clinical risk models. Ann Thorac Surg 2001;72:323– 6. DeLong ER, DeLong DM, Clark-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves; a nonparametric approach. Biometrics 1988;44:837– 45. Cerfolio RJ, Ojha B, Mukherjee S, et al. Positron emission tomography scanning with 2-fluoro-2-deoxy-d-glucose as a predictor of response of neoadjuvant treatment for nonsmall cell carcinoma. J Thorac Cardiovasc Surg 2003;125(4): 938 – 44. Cerfolio RJ, Buddhiwardhan O, Bryant AS, et al. The role of FDG-PET scan in staging patients with non-small cell carcinoma. Ann Thorac Surg 2003;76:861– 6. Gupta NC, Graeber GM, Bishop HA, et al. Comparative efficacy of positron emission tomography with fluorodeoxyglucose in evaluation of small, intermediate, and large lymph node lesions. Chest 2000;117:773– 8. Roberts PF, Follette DM, von Haag D, et al. Factors associated with false-positive staging of lung cancer by positron emission tomography. Ann Thorac Surg 2000;70:1154 – 60. Pieterman RM, van Putten JW, Meuzelaar JJ, et al. Preoperative staging of non-small cell lung cancer with positronemission tomography. N Engl J Med 2000;343:254 – 61. Vansteenkiste JF, Stroobants SG, Dupont PJ, et al. Prognostic importance of the standardized uptake value on (18)F-fluoro-2-deoxy-glucose-positron emission tomography scan in non-small-cell lung cancer: an analysis of 125 cases. Leuven Lung Cancer Group. J Clin Oncol 1999; 17(10):3201– 6. Galetta D, Cesario A, Margaritora S, et al. Enduring challenge in the treatment of nonsmall cell lung cancer with clinical stage IIIB: results of a trimodality approach. Ann Thorac Surg 2003;76:1802–9. Rusch VW, Giroux DJ, Kraut MJ, et al. Induction chemoradiation and surgical resection for non-small cell lung carcinomas of the superior sulcus: initial results of Southwest Oncology Group Trial 9416 (Intergroup Trial 0160). J Thorac Cardiovasc Surg 2001;121(3):472– 83.
24. Ryu J, Choi NC, Fischman AJ, et al. FDG-PET in staging and restaging non-small cell lung cancer after neoadjuvant chemoradiotherapy: correlation with histopathology. Lung Cancer 2002;35:179 – 87. 25. Akhurst T, Downey RJ, Ginsberg MS, et al. An initial experience with FDG-PET in the imaging of residual disease
CERFOLIO ET AL REPEAT FDG-PET AFTER NEOADJUVANT THERAPY
1909
after induction therapy for lung cancer. Ann Thor Surg 2002;73(1):259 – 64. 26. Haberkorn U, Strauss LG, Dimitrakopoulou A, et al. PET studies of fluorodeoxyglucose metabolism in patients with recurrent colorectal tumors receiving radiotherapy. J Nucl Med 1991;32(8):1485–90.
INVITED COMMENTARY The article by Cerfolio and colleagues is an important addition to the literature in regard to our continuing struggle with how to best incorporate multidisciplinary care in the management of lung cancer patients. The fact that this is necessary is no longer debatable in this author’s opinion. More than 70% of patients have systemic disease at the time of diagnosis, and we have hopefully and finally come to grips with the fact that a recurrence rate of 20% to 40% for “surgically curable” patients is not satisfactory. It has been known for quite some time by our medical oncology colleagues that radiographic findings do not necessarily correlate with chemotherapy effects on a solid tumor, and some of these clinicians, such as those that treat gastrointestinal stromal and germ cell tumors, have already been using positron emission tomography (PET) and computed tomography (CT) as a measure of true response. However, how to apply this to the treatment of lung cancer is a bit more problematic. After years of equivocal “statistical near miss” reports and a great deal of debate, recent more definitive studies from Europe have suggested that adjuvant chemotherapy is probably as important in lung cancer patients, regardless of stage, as it has been for those with breast cancer. Although at many centers the use of neoadjuvant treatment for stage IIIA has been common, it has not been routine practice around the world, and with this new information many centers have now abandoned its use for adjuvant protocols. The Intergroup S9900 trial utilizing neoadjuvant chemotherapy for early stage lung cancer patients has been closed due in large part to this shift of thinking and associated lack of recent accrual, and the final data analysis is pending. Where does this leave the use of PET and CT in the evaluation of response to treatment for lung cancer being considered as candidates for surgical resection? It is
© 2004 by The Society of Thoracic Surgeons Published by Elsevier Inc
likely that some centers will continue to use neoadjuvant treatment for stage IIIA patients, and many will continue its use in T4 tumors that are considered to be “borderline” resectable. Certainly it could be recommended that we continue to utilize PET and CT scanning in these patients to aid in the decision making process for whom to resect after these treatments. As we continue to develop more effective biologically based systemic agents as well as more sensitive imaging techniques, the use of PET and CT and related modalities may well be expanded. In these scenarios, the decision to operate on localized disease may hinge less on the continued presence of a radiographic density than on an abnormality that demonstrates metabolic activity, or “signs of life.” Patients with metastatic disease may also be impacted, as these types of imaging are used to determine more accurate response at distant sites. We may begin to expand surgical treatment for those that have eradication of metastatic foci, and offer subsequent “local control” or “salvage” surgical therapy for the primary lesion. In either case, metabolic imaging is likely to be an increasingly useful adjunct to our decision making process for surgical treatment of thoracic malignancies. W. Roy Smythe, MD Department of Surgery Texas A&M University Health Science Center College of Medicine Scott and White Memorial Hospital 2401 S 31st St Temple, TX 76508 e-mail: [email protected]
0003-4975/04/$30.00 doi:10.1016/j.athoracsur.2004.07.069
GENERAL THORACIC
Ann Thorac Surg 2004;78:1903–9
Background. Repeat positron emission tomography (PET) with 18F-fluorodeoxyglucose (FDG) and chest computed tomography (CT) are used to assess the effectiveness of chemoradiotherapy in patients with non-small cell lung cancer (NSCLC); however, the change in the standardized uptake values (SUV) has not been correlated with the pathologic change of the primary tumor. Methods. This is a retrospective cohort study of a prospective database of 56 patients who had NSCLC, FDG-PET, and chest CT scans both before and after neoadjuvant therapy, followed by complete resection of their cancer. Maximum SUVs (maxSUV) and tumor size were measured, and the percentage of change was compared with the percentage of nonviable tumor cells. The primary objective was to measure the degree of correlation between these values. Results. The change in the maxSUV has a near linear
relationship to the percent of nonviable tumor cells in the resected tumors. FDG-PET’s maxSUV is better correlated to pathology than the change in size on CT scan (r2 ⴝ 0.75, r2 ⴝ 0.03, p < 0.001). When the maxSUV decreased by 80% or more, a complete pathologic response could be predicted with a sensitivity of 90%, specificity of 100%, and accuracy of 96%. Conclusions. The change in maxSUV on FDG-PET scan after neoadjuvant therapy holds a near linear relationship with pathologic response. It is a more accurate predictor than the change of size on CT scan. When the maxSUV decreases by 80% or more it is likely that the patient is a complete responder irrespective of cell type, neoadjuvant treatment, or the final absolute maxSUV. These findings may help guide treatment strategies. (Ann Thorac Surg 2004;78:1903–9) © 2004 by The Society of Thoracic Surgeons
L
that undergo resection have a significant survival advantage [11]. Thus, being able to identify responders leads to improved patient selection for surgery and may also help guide further treatment for patients with stage IV disease. Chest computed tomography (CT) and positron emission tomography (PET) using 18F-fluorodeoxyglucose (FDG) are most often used to assess the response of NSCLC to treatment. However little data have shown that the clinical response rate detected on repeat FDGPET scan actually correlates with the pathologic response of the primary tumor [12]. Therefore, we evaluated the effectiveness of repeat FDG-PET and repeat CT scan as predictors of this response in a group of patients who met strict entry criteria.
ung cancer is the leading cause of cancer deaths worldwide. More Americans die from lung cancer than the next three most common solid organ cancers (breast, prostate, and colon) combined [1– 4]. Because of these dismal numbers, more patients are undergoing neoadjuvant therapy before resection, adjuvant therapy after resection, and continued palliative chemotherapy. Several reports of prospective studies have shown increased survival favoring neoadjuvant therapy for patients with stage Ib, IIa, and N2 (stage IIIa) disease [5–9]. Patients with N2 disease who have had neoadjuvant therapy are usually only considered for resections if the N2 disease resolves. Similarly, further chemotherapy or changes in chemotherapy are often made for those patients with stage IV disease based on response. Improved survival has been shown in patients who have a response, even with stage IV disease [10]. Similarly, patients who are complete responders
Accepted for publication June 2, 2004. Address reprint requests to Dr Cerfolio, Division of Cardiothoracic Surgery, University of Alabama at Birmingham, 1900 University Blvd, THT 712, Birmingham, AL 35294; e-mail: [email protected].
© 2004 by The Society of Thoracic Surgeons Published by Elsevier Inc
Material and Methods Patients Over a 24-month period (January 2002 until December 2003), one general thoracic surgeon (RJC) performed 1,963 operations at the University of Alabama at Birmingham (UAB). Ninety-four patients had neoadjuvant therapy before resection, but only 56 met the strict 0003-4975/04/$30.00 doi:10.1016/j.athoracsur.2004.06.102
GENERAL THORACIC
Repeat FDG-PET After Neoadjuvant Therapy is a Predictor of Pathologic Response in Patients With Non-Small Cell Lung Cancer
GENERAL THORACIC
1904
CERFOLIO ET AL REPEAT FDG-PET AFTER NEOADJUVANT THERAPY
Ann Thorac Surg 2004;78:1903–9
Table 1. Entry Criteria
Table 2. Patient Characteristics
Patients were required to have the following: 1. Biopsy-proven NSCLC 2. An initial CT scan with 5 mm contiguous columnated slices with intravenous contrast prior to the start of neoadjuvant therapy. 3. An initial FDG-PET scan (on a dedicated FDG-PET scanner or an integrated PET-CT scanner) with the maxSUV measured prior to the start of neoadjuvant therapy. 4. The initial FDG-PET and chest CT scans had to be performed within 1 month prior to the start of the neoadjuvant therapy 5. Patients received neoadjuvant therapy (either using chemotherapy alone or combined chemoradiotherapy). 6. A repeat FDG-PET scan using the same PET center within 1 month of the completion of the neoadjuvant therapy using an identical dose of FDG with similar scanning techniques and with the maxSUV’s reported. 7. A repeat chest CT scan of the chest with 5 mm columnated cuts with intravenous contrast within 1 month of the completion of the neoadjuvant therapy. 8. A posterior-lateral thoracotomy with complete resection of the primary tumor along with complete thoracic lymphadenectomy within 5 weeks of the completion of the neo-adjuvant therapy.
Characteristics
CT ⫽ computed tomography; FDG-PET ⫽ 18F-fluorodeoxyglucose positron emission tomography; maxSUV ⫽ maximum standardized uptake values; NSCLC ⫽ non-small cell lung cancer.
entry criteria described in Table 1. The most common reason for exclusion of the other 38 patients was the lack of a dedicated FDG-PET scan or the lack of a maximum standardized uptake value (maxSUV) reported before the start of their neoadjuvant therapy. The remaining 56 patients comprised the cohort for this study. UAB’s Institutional Review Board approved this review of our prospective database. Patients who received neoadjuvant chemotherapy under the guise of the Southern Western Oncology Group S9900 trial received an additional separate consent. Patients were excluded if they had a history of type I diabetes mellitus, had a FDG-PET preformed on a nondedicated camera, did not have the maxSUV reported, had a delay greater than 6 weeks from the completion of their neoadjuvant treatment to resection, had an incomplete resection, were medically unfit, or refused surgery.
Imaging The PET scans performed at UAB were conducted on an integrated PET-CT scanner (GE Discovery LS PET-CT Scanner, Milwaukee, WI). Patients were asked to fast for 4 hours and then subsequently received 555 MBq (15 mCi) of FDG intravenously followed by PET after 1 hour. The maxSUV was determined by drawing regions of interest (ROI) on the attenuation-corrected FDG-PET images around the primary tumor. It was then calculated by the software contained within the PET-CT scanner by the formula: [13]
N ⫽ 56
Median age (range) Gender Male Female Neoadjuvant treatment type Chemoradiotherapy Chemotherapy only Median duration between last neoadjuvant treatment and surgery (range) Tumor histology Adenocarcinoma Squamous cell Other Preoperative cancer stage Ib–T2N0 IIa/IIb–T1, T2N1 IIb–T3N0 IIIa–T1, T2, T3N2 IIIa–T3N1 Stage IV–T2N0M1 (brain met) Type of operation Lobectomy Segmentectomy Pneumonectomy
63 (42–76) 31 (55%) 25 (45%) 23 (41%) 33 (59%) 29.8 days (20–40 days) 28 (50%) 22 (39%) 6 (11%) 11 (20%) 9 (16%) 8 (14%) 23 (41%) 4 (7%) 1 (2%) 43 (77%) 8 (14%) 5 (9%)
C (Ci ⁄ mL) MaxSUV ⫽
ID (Ci) w (kg)
where C ⫽ activity at a pixel within the tissue defined by an ROI and ID ⫽ injected dose per kg of the patient’s body weight (w). The maxSUV within the selected ROI’s was used. The percentage of change of the maxSUV was calculated by the formula: maxSUV initial ⫺ maxSUV final max SUV initial Patients that had FDG-PET scans performed at outside hospitals were included in this trial if the FDG-PET was performed on a dedicated PET camera with bismuth germanate crystals. All patients were carefully staged and biopsy specimens were taken from all suspicious N2, N3, or M1 areas (maxSUV ⬎ 2.5). Patients with N2 disease had chemoradiotherapy, and those that were N2 negative had chemotherapy alone. Biopsies were performed on all suspicious M1 metastatic lesions unless cancer was suspected in the bone or brain where magnetic resonance imaging (MRI) was considered to be the gold standard. Patients with suspected M1 disease in the liver, adrenal, or contralateral lung underwent definitive biopsy to prove or disprove M1 cancer. The volume of the tumor on chest CT was calculated using the equation:
CERFOLIO ET AL REPEAT FDG-PET AFTER NEOADJUVANT THERAPY
1905
Fig 1. Correlation between the decrease in maximum standardized uptake value (maxSUV) and the percentage of nonviable tumor in the resected specimens. The regression line equation: y ⫽ 0.9501 ⫻ ⫹5.230; r2 ⫽ 0.75, p ⬍ 0.001. Values where more than one data point was observed have been indicated by a solid black box and the number of observations are indicated in parenthesis. The open boxes represent one data point. Dashed lines indicate the 95% confidence interval for the regression line (solid line).
4 Volume ⫽ abc 3 with a, b, and c representing the height, diameter, and width of the mass (cm), respectively. This volume was calculated on the initial chest CT and then on the repeat chest CT after the neoadjuvant therapy was completed. The percentage of change in size was calculated by the formula: volume initial 共cm3) ⫺ volume final 共cm3) volume initial 共cm3)
Surgery Thoracotomy with complete thoracic lymphadenectomy was performed in all patients in this study. Patients who initially had stage IIIa disease, secondary to N2 disease, in general were offered pulmonary resection only if the initial nodal station involved with cancer was negative for cancer on repeat biopsy after the neoadjuvant therapy was completed. Repeat transesophageal ultrasound with fine needle aspirate was used to reassess lymph node stations 7, 8, and 9, and sometimes 5. Repeat videoassisted thorascopy was used to perform a repeat biopsy of stations 5 and 6. Repeat biopsies were performed on right-sided stations 2 and 4 and left-sided 4 by using open thoracotomy and frozen section analysis. Repeat mediastinoscopy was not performed. Segmentectomy, lobectomy, or pneumonectomy was performed; and wedge resection was not. The bronchial stump was buttressed with an intercostal muscle flap in all patients. As part of the entry criteria for this study, all patients had complete resection with negative margins and complete thoracic lymphadenectomy.
Pathologic Analysis Multiple hematoxylin and eosin stained sections of each tumor were reviewed by a pathologist and then rereviewed by another pathologist (TSW) who carefully calculated the percentage of nonviable tumor after recutting the entire tumor and scrutinizing multiple cross-
sections. The percent of nonviable tumor was defined as the combined percentage of scar and necrosis. Scar was defined as fibrous tissue intimately admixed with residual tumor. Patients with mixed tumors were labeled as squamous cell or adenocarcinoma based on the predominant cellular type seen. Complete pathologic response was defined as 1% or less of viable tumor cells detected on pathologic review of the entire resected specimens. The pathologists were blinded to all clinical, radiologic, and surgical findings.
Statistical Analysis The primary outcome evaluated was the degree of correlation between the percentage of change in maxSUV and the percentage of nonviable tumor in the resected specimens. We also evaluated the degree of correlation between the percentages of decrease of the volume of the mass on repeat CT scan and of nonviable tumor in the resected specimen. In addition, three secondary outcomes were also evaluated. The first was to assess the ability of FDG-PET and CT scan to predict complete pathologic response. The second was to compare the degree of correlation based on the type of neoadjuvant therapy. The third outcome assessed was to compare the degree of correlation based on the two main tissue types of cancer seen in this study, squamous cell and adenocarcinoma. Spearman’s rank-correlation coefficient was used to determine the relationship and establish a linear equation for estimating the percentage of nonviability of tumor according to the percentage of change in the maxSUV on FDGPET and also the change in volume of the mass on CT scan. Partial correlation coefficients were used to investigate relationships while adjusting for covariates. Multiple linear regression step-wise analysis was used to adjust for risk factors and identify any variables that were independently associated with a decrease in maxSUV (age, gender, percentage of nonviability of tumor, dose of radiation, cell type, and type of neoadjuvant therapy). The highest combined sensitivity and specificity values generated from the receiver operating characteristic (ROC) curves that predicted
GENERAL THORACIC
Ann Thorac Surg 2004;78:1903–9
GENERAL THORACIC
1906
CERFOLIO ET AL REPEAT FDG-PET AFTER NEOADJUVANT THERAPY
Ann Thorac Surg 2004;78:1903–9
Table 3. Actual Data Used to Compute Sensitivity, Specificity and Accuracy Values for Complete Responders for Decrease in Maximum Standardized Uptake Values and in Mass Volume on Computed Tomography FDG-PET maxSUV decrease
True Positive
False Positive
False Negative
True Negative
19 18 17 12
2 1 0 0
0 1 2 7
35 36 37 37
ⱖ60% ⱖ70% ⱖ80% ⱖ90% Decrease in volume on CT
Primary Outcomes
True Positive
False Positive
False Negative
True Negative
12 9 9 8
17 15 14 12
7 10 10 11
20 22 23 25
ⱖ60% ⱖ70% ⱖ80% ⱖ90%
CT ⫽ computed tomography; positron emission tomography; uptake values.
5 other patients without N2 disease received chemotherapy alone at outside institutions. Three of these patients had carboplatin and paclitaxel. Twenty-three patients with N2 disease underwent combined chemoradiotherapy, 18 of whom had carboplatin and paclitaxel. The dose of radiotherapy used was 45 to 51 Gy in 8 patients and 60 to 68 Gy in 15 patients. The median dose of preoperative radiation was 60 Gy (range 45 to 68 Gy). There was one operative death (1.8%) and no bronchopleural fistulas.
FDG-PET ⫽ 18F-fluorodeoxyglucose maxSUV ⫽ maximum standardized
complete pathologic responders were identified [14]. The ROC curves for FDG-PET and CT were compared according to the method of DeLong and associates [15]. The test for proportions and the binomial approximation test were used to compare the percent accuracy values between FDG-PET and CT scan. A two-sided p value of 0.05 or less was considered to indicate statistical significance. SAS version 9.0 (SAS Institute, Cary, NC) was used to conduct this analysis.
Results Patient Characteristics There were 56 patients (31 men) with a median age of 63 (range 42 to 76). Patient characteristics are shown in Table 2. Thirty-three patients had chemotherapy alone, and 23 patients with stage Ib, IIa, or non-N2 IIIa were enrolled in the Southern Western Oncology Group S9900 trial. All received carboplatin and paclitaxel. In addition,
Figure 1 demonstrates the percentage of change in the maxSUV plotted against the percentage of nonviable tumor for all patients. A near linear relationship is demonstrated (r2 ⫽ 0.75, p ⬍ 0.001). The correlation was stronger after adjusting for cell type, radiation dose, and type of neoadjuvant therapy (r2 ⫽ 0.81, p ⬍ 0.001). Age, gender, surgery type, and duration between the last dose of neoadjuvant therapy and resection were not significantly associated with the percentage of decrease in maxSUV. No correlation was evident between the change in tumor size on repeat CT scan and the percentage of nonviable tumor (r2 ⫽ 0.03, p ⬎ 0.05).
Secondary Outcomes/Complete Responders Nineteen patients were complete responders. Seven of the 33 patients (21%) who had chemotherapy achieved a complete response rate, whereas 12 of the 23 patients (52%) who had combined chemoradiotherapy had a complete pathologic response rate. The sensitivity and specificity of various cut points in the percentage of change of maxSUV and change in volume on CT scan were determined and are shown in Tables 3 and 4. We found that FDG-PET is superior to CT scan at all cut points for all variables (p ⬍ 0.05 for all). In addition, these variables are maximized when a cut point of 80% or greater is selected for the change in the maxSUV. Figure 2 shows the ROC curves that plot the sensitivity and specificity of the complete responders, and it again shows the superiority of FDG-PET (p ⬍ 0.001) over CT scan. In addition, we found no difference in the accuracy of the percentage of change of maxSUV for patients with N2 disease when compared with those without.
Table 4. The Sensitivity, Specificity and Accuracy of Predicting Complete Pathologic Response by the Percentage of Decrease in Maximum Standardized Uptake Values on Repeat FDG-PET and the Percentage of Decrease in the Mass Size of Tumor on CT Scana Sensitivity
ⱖ60% ⱖ70% ⱖ80% ⱖ90% a
Specificity
Accuracy
PPV
NPV
FDG-PET
CT
FDG-PET
CT
FDG-PET
CT
FDG-PET
CT
FDG-PET
CT
100% 95% 90% 63%
63% 47% 47% 42%
95% 97% 100% 100%
54% 59% 63% 68%
96% 96% 96% 88%
57% 56% 57% 59%
90% 94% 100% 100%
41% 38% 27% 40%
100% 97% 95% 84%
74% 69% 70% 69%
All p values are ⬍0.05.
FDG-PET ⫽ 18F-fluorodeoxyglucose positron emission tomography;
NPV ⫽ negative predictive value;
PPV ⫽ positive predictive value.
Fig 2. Receiver operating characteristic curves of sensitivity (solid lines) plotted against 100 minus specificity for 18F-fluorodeoxyglucose positron emission tomography (FDG-PET) and computed tomography (CT) for complete responders. (Area under curve ⫽ 0.935 for FDG-PET and 0.532 for CT, p ⬍ 0.001).
Type of Neoadjuvant Therapy FDG-PET was better able to predict the pathologic response in patients who had chemotherapy alone than in those that had chemoradiotherapy (r2 ⫽ 0.76 and 0.39, respectively; p ⬍ 0.001). Additionally, we observed a difference when we compared low-dose (ⱕ 51 Gy) radiation with high-dose (ⱖ 60 Gy) radiation (r2 ⫽ 0.90 and 0.66, respectively; p ⬍ 0.001). Repeat FDG-PET scan was a better predictor of pathologic response when lower doses of radiation were used.
Cell Type We found that FDG-PET was better able to predict the pathologic response in patients who had squamous cell cancer than in those that had adenocarcinoma (r2 ⫽ 0.83 and 0.58, respectively; p ⬍ 0.001). Interestingly, we also found that those with squamous cell cancer had a better response to their neoadjuvant therapy. The median decrease in maxSUV was 80% in patients with squamous cell cancer, and the median nonviable tumor found on pathologic examination was 88%. In contrast, the median decrease in maxSUV in patients with adenocarcinoma was 60%, and the median nonviable tumor found on pathologic examination for adenocarcinoma was 69%.
Comment Many studies have demonstrated the problem of falsepositive and false-negative results with FDG-PET scans in patients with NSCLC and the problems of restaging [16 –24]. One recurring theme is that FDG-PET scans do
CERFOLIO ET AL REPEAT FDG-PET AFTER NEOADJUVANT THERAPY
1907
not supplant the need for tissue biopsy specimens. We, along with others, have shown that a repeat FDG-PET scan may accurately predict the status of lymph nodes after neoadjuvant therapy [16, 24]; however, few reports have shown that repeat FDG-PET is an accurate predictor of the pathologic response of the primary tumor [12, 24, 25]. Nevertheless, it is commonly ordered, and the results are frequently used to guide, change, or even deny subsequent chemoradiotherapy or surgery, or both. This trial featured stringent entry criteria on a consecutive series of patients and showed that the change in maxSUV on a repeat FDG-PET scan is an accurate predictor of the pathologic response. The strength of this study, compared with other reports, is the entry criteria applied. For example, Ryu and colleagues in 2002 [24] retrospectively studied 26 patients and found that the change in SUVs of the primary tumor was useful for monitoring the therapeutic effect of neoadjuvant chemoradiotherapy in patients with NSCLC. Akhurst and colleagues in 2002 [25] reported a retrospective series on 56 patients and found that FDGPET after induction therapy accurately detected residual viable primary tumor. However, those reports did not require a baseline FDG-PET to show the percentage of change in maxSUV, a dedicated FDG-PET scan, or a separate tumor type analysis, as was done in our study. A limitation of our study is the short follow-up period and thus the lack of survival data. Yet others have shown that a significant response rate does equate to improved survival [10, 11]. Further studies with long-term analysis are required. The ability to noninvasively identify complete responders to neoadjuvant therapy could have important implications for surgical selection. Pisters and colleagues showed in 1993 [11] that the 5-year survival of patients who underwent resection and who had a complete pathologic response was 54% compared with 15% in those without a complete pathologic response. Nineteen (35%) patients in the study were complete responders, a similar incidence reported by Ryu and colleagues [24]. We found that when the maxSUV decreases by more than 80%, a complete responder can be predicted with 96% accuracy. In contrast, Port and colleagues [12] reported in 2004 that repeat PET did not predict the response to neoadjuvant therapy. However, in that report the authors arbitrarily defined a major PET response as a “reduction in the SUV of 50% or more.” If they had chosen a percentage of change of 80% or greater in maxSUV instead of 50% for their data they would have found that repeat PET correctly predicted 4 out of the 5 complete responders in their series. Importantly, we have demonstrated that the maxSUV does not need to fall to 0 to imply a complete responder, but rather it needs to have an 80% or greater reduction. In those patients who had preoperative chemotherapy, 7 patients were complete responders and in 5, the repeat maxSUV fell to 0. However, in the 12 complete responders who received combined chemoradiotherapy, only 6 had a maxSUV of 0 on the repeat FDG-PET. Thus, a patient who is a complete responder may have a maxSUV on the repeat FDG-PET greater than 0, especially if
GENERAL THORACIC
Ann Thorac Surg 2004;78:1903–9
GENERAL THORACIC
1908
CERFOLIO ET AL REPEAT FDG-PET AFTER NEOADJUVANT THERAPY
the patient had preoperative radiation. The percentage of drop is the best predictor of complete pathologic response and not the absolute value of the maxSUV on the repeat FDG-PET scan. One potential concern is that the SUV measured on one PET scanner may be different on another. The term standardized uptake value used to be an oxymoron because it was anything but standardized. However, most all dedicated PET scanners now use software packages that automatically calculate the maxSUV and take into account many of the various factors that can affect its value. Moreover, protocols for PET scanning are become more standardized, and centers are starting to use similar techniques of scanning, dosages of FDG, time intervals from injection to scanning, and baseline glucose, among others. These steps will make SUVs standardized across centers and continents. To protect for this potential problem in this study, we required that the patients received both FDG-PET scans at the same center using similar techniques. Finally, we chose to look at maxSUV instead of mean SUV because it eliminates many of the subjective measurements inherent to the latter. Interestingly we found that FDG-PET was less accurate for determining the pathologic response after neoadjuvant therapy when patients had radiation in addition to chemotherapy as opposed to chemotherapy alone. Other studies have demonstrated that radiation can interfere with the interpretation of FDG-PET scans [26]. We also found that the FDG-PET after neoadjuvant therapy was a more accurate predictor of response in patients who had squamous cell cancers than those who had adenocarcinoma, but the reasons for this are unclear. Our report shows that the percentage of change in the maxSUV on FDG-PET scan after neoadjuvant treatment is an accurate predictor of the actual pathologic response of the primary tumor in patients with NSCLC and that it also can identify complete responders. It is more accurate than CT scan. This information may help guide subsequent treatment strategies for medical oncologists, radiation oncologists, and surgeons. Although, this study and our conclusions must obviously be confined to patients with NSCLC, some of its findings may be applicable to patients with other types of solid organ cancers, especially ones that feature squamous cell or adenocarcinoma. Further prospective trials with long-term follow-up are needed.
References 1. Weir HK, Thun MJ, Hankey BF, et al. Annual report to the nation on the status of cancer, 1975–2000, featuring the uses of surveillance data for cancer prevention. J Nat Cancer Ins 2003;95(17):1276 –99. 2. Jemal A, Murray T, Samuels A, et al. Cancer statistics, 2003. CA Cancer J Clin 2003;53:5–26. 3. Parkin DM, Pisani P, Ferlay J. Estimates of the worldwide incidence of eighteen major cancers in 1985. Int J Cancer 1993;54:594 – 606. 4. Spira A, Ettinger DS. Multidisciplinary management of lung cancer. N Engl J Med 2004;3550:379 –92. 5. Depierre A, Milleron B, Moro-Sibilot, D, et al. Preoperative chemotherapy followed by surgery compared with primary
Ann Thorac Surg 2004;78:1903–9
6.
7.
8.
9.
10.
11.
12.
13. 14. 15.
16.
17. 18.
19. 20. 21.
22.
23.
surgery in resectable stage 1 (except T1N0), II and IIIa non-small cell lung cancer. J Clin Oncol 2002;20:247–53. Roth JA, Fossella F, Komaki R, et al. A randomized trial comparing perioperative chemotherapy and surgery with surgery alone in resectable stage IIIA non-small-cell lung cancer. J Natl Cancer Inst 1994;86(9):673– 80. Roth JA, Atkinson EN, Fossella F, et al. Long-term follow-up of patients enrolled in a randomized trial comparing perioperative chemotherapy and surgery with surgery alone in resectable stage IIIA non-small-cell lung cancer. Lung Cancer 1998;21(1):1– 6. Rosell R, Gomez-Codina J, Camps C, et al. Preresectional chemotherapy in stage IIIA non-small-cell lung cancer: a 7-year assessment of a randomized controlled trial. Lung Cancer 1999;26(1):7–14. Rosell R, Gomez-Codina J, Camps C, et al. A randomized trial comparing preoperative chemotherapy plus surgery with surgery alone in patients with non-small-cell lung cancer. N Engl J Med 1994;330:153– 8. Weber WA, Petersen V, Schmidt B, et al. Positron emission tomography in non-small-cell lung cancer: prediction of response to chemotherapy by quantitative assessment of glucose use. J Clin Oncol 2003;21(14):2651–7. Pisters KM, Ginsberg RJ, Giroux DJ, et al. Induction chemotherapy before surgery for early-stage lung cancer: a novel approach. Bimodality Lung Oncology Team. J Thorac Cardiovasc Surg 2000;119(3):429 –39. Port JL, Kent MS, Korst RJ, Keresztes R, Levin MA, Altorki NK. Positron emission tomography scanning poorly predicts response to preoperative chemotherapy in non-small cell lung cancer. Ann Thorac Surg 2004;77(1):254 –9. Nabi HA, Zubeldia JM. Clinical applications of F18-FDG in oncology. J Nucl Med Technol 2002;30:3–9. Grunkemeier GL, Jin R. Receiver operating characteristic curve analysis of clinical risk models. Ann Thorac Surg 2001;72:323– 6. DeLong ER, DeLong DM, Clark-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves; a nonparametric approach. Biometrics 1988;44:837– 45. Cerfolio RJ, Ojha B, Mukherjee S, et al. Positron emission tomography scanning with 2-fluoro-2-deoxy-d-glucose as a predictor of response of neoadjuvant treatment for nonsmall cell carcinoma. J Thorac Cardiovasc Surg 2003;125(4): 938 – 44. Cerfolio RJ, Buddhiwardhan O, Bryant AS, et al. The role of FDG-PET scan in staging patients with non-small cell carcinoma. Ann Thorac Surg 2003;76:861– 6. Gupta NC, Graeber GM, Bishop HA, et al. Comparative efficacy of positron emission tomography with fluorodeoxyglucose in evaluation of small, intermediate, and large lymph node lesions. Chest 2000;117:773– 8. Roberts PF, Follette DM, von Haag D, et al. Factors associated with false-positive staging of lung cancer by positron emission tomography. Ann Thorac Surg 2000;70:1154 – 60. Pieterman RM, van Putten JW, Meuzelaar JJ, et al. Preoperative staging of non-small cell lung cancer with positronemission tomography. N Engl J Med 2000;343:254 – 61. Vansteenkiste JF, Stroobants SG, Dupont PJ, et al. Prognostic importance of the standardized uptake value on (18)F-fluoro-2-deoxy-glucose-positron emission tomography scan in non-small-cell lung cancer: an analysis of 125 cases. Leuven Lung Cancer Group. J Clin Oncol 1999; 17(10):3201– 6. Galetta D, Cesario A, Margaritora S, et al. Enduring challenge in the treatment of nonsmall cell lung cancer with clinical stage IIIB: results of a trimodality approach. Ann Thorac Surg 2003;76:1802–9. Rusch VW, Giroux DJ, Kraut MJ, et al. Induction chemoradiation and surgical resection for non-small cell lung carcinomas of the superior sulcus: initial results of Southwest Oncology Group Trial 9416 (Intergroup Trial 0160). J Thorac Cardiovasc Surg 2001;121(3):472– 83.
24. Ryu J, Choi NC, Fischman AJ, et al. FDG-PET in staging and restaging non-small cell lung cancer after neoadjuvant chemoradiotherapy: correlation with histopathology. Lung Cancer 2002;35:179 – 87. 25. Akhurst T, Downey RJ, Ginsberg MS, et al. An initial experience with FDG-PET in the imaging of residual disease
CERFOLIO ET AL REPEAT FDG-PET AFTER NEOADJUVANT THERAPY
1909
after induction therapy for lung cancer. Ann Thor Surg 2002;73(1):259 – 64. 26. Haberkorn U, Strauss LG, Dimitrakopoulou A, et al. PET studies of fluorodeoxyglucose metabolism in patients with recurrent colorectal tumors receiving radiotherapy. J Nucl Med 1991;32(8):1485–90.
INVITED COMMENTARY The article by Cerfolio and colleagues is an important addition to the literature in regard to our continuing struggle with how to best incorporate multidisciplinary care in the management of lung cancer patients. The fact that this is necessary is no longer debatable in this author’s opinion. More than 70% of patients have systemic disease at the time of diagnosis, and we have hopefully and finally come to grips with the fact that a recurrence rate of 20% to 40% for “surgically curable” patients is not satisfactory. It has been known for quite some time by our medical oncology colleagues that radiographic findings do not necessarily correlate with chemotherapy effects on a solid tumor, and some of these clinicians, such as those that treat gastrointestinal stromal and germ cell tumors, have already been using positron emission tomography (PET) and computed tomography (CT) as a measure of true response. However, how to apply this to the treatment of lung cancer is a bit more problematic. After years of equivocal “statistical near miss” reports and a great deal of debate, recent more definitive studies from Europe have suggested that adjuvant chemotherapy is probably as important in lung cancer patients, regardless of stage, as it has been for those with breast cancer. Although at many centers the use of neoadjuvant treatment for stage IIIA has been common, it has not been routine practice around the world, and with this new information many centers have now abandoned its use for adjuvant protocols. The Intergroup S9900 trial utilizing neoadjuvant chemotherapy for early stage lung cancer patients has been closed due in large part to this shift of thinking and associated lack of recent accrual, and the final data analysis is pending. Where does this leave the use of PET and CT in the evaluation of response to treatment for lung cancer being considered as candidates for surgical resection? It is
© 2004 by The Society of Thoracic Surgeons Published by Elsevier Inc
likely that some centers will continue to use neoadjuvant treatment for stage IIIA patients, and many will continue its use in T4 tumors that are considered to be “borderline” resectable. Certainly it could be recommended that we continue to utilize PET and CT scanning in these patients to aid in the decision making process for whom to resect after these treatments. As we continue to develop more effective biologically based systemic agents as well as more sensitive imaging techniques, the use of PET and CT and related modalities may well be expanded. In these scenarios, the decision to operate on localized disease may hinge less on the continued presence of a radiographic density than on an abnormality that demonstrates metabolic activity, or “signs of life.” Patients with metastatic disease may also be impacted, as these types of imaging are used to determine more accurate response at distant sites. We may begin to expand surgical treatment for those that have eradication of metastatic foci, and offer subsequent “local control” or “salvage” surgical therapy for the primary lesion. In either case, metabolic imaging is likely to be an increasingly useful adjunct to our decision making process for surgical treatment of thoracic malignancies. W. Roy Smythe, MD Department of Surgery Texas A&M University Health Science Center College of Medicine Scott and White Memorial Hospital 2401 S 31st St Temple, TX 76508 e-mail: [email protected]
0003-4975/04/$30.00 doi:10.1016/j.athoracsur.2004.07.069
GENERAL THORACIC
Ann Thorac Surg 2004;78:1903–9