neoadjuvant therapy of esophageal squamous cell carcinoma: response evaluation by positron emission...

Neoadjuvant Therapy of Esophageal SquamousCell Carcinoma: Response Evaluation by PositronEmission TomographyBjörn L.D.M. Brücher, MD,* Wolfgang Weber, MD,† Markus Bauer, MD,‡ Ullrich Fink, MD,*Norbert Avril, MD,† Hubert J. Stein, MD,* Martin Werner, MD,‡ Frank Zimmerman, MD,§J. Rüdiger Siewert, MD, FACS (Hon), FRCS (Edinb)* and Markus Schwaiger, MD,†

From the *Chirurgische Klinik und Poliklinik, †Nuklearmedizinische Klinik und Poliklinik, ‡Institut für Allgemeine Pathologie undPathologische Anatomie, and the §Klinik und Poliklinik für Strahlentherapie und Radiologische Onkologie, Klinikum rechts derIsar der Technischen Universität, Munich, Germany

ObjectiveTo evaluate the use of positron emission tomography using[18F]-fluorodeoxyglucose (FDG-PET) to assess the responseto neoadjuvant radiotherapy and chemotherapy in patientswith locally advanced esophageal cancer.

Summary Background DataImaging modalities, including endoscopy, endoscopic ultra-sound, computed tomography, and magnetic resonance im-aging, currently used to evaluate response to neoadjuvanttreatment in esophageal cancer do not reliably differentiatebetween responders and nonresponders.

MethodsTwenty-seven patients with histopathologically proven squa-mous cell carcinoma of the esophagus, located at or abovethe tracheal bifurcation, underwent neoadjuvant therapy con-sisting of external-beam radiotherapy and 5-fluorouracil as acontinuous infusion. FDG-PET was performed before and 3weeks after the end of radiotherapy and chemotherapy (be-

fore surgery). Quantitative measurements of tumor FDG up-take were correlated with histopathologic response and pa-tient survival.

ResultsAfter neoadjuvant therapy, 24 patients underwent surgery.Histopathologic evaluation revealed less than 10% viable tu-mor cells in 13 patients (responders) and more than 10% via-ble tumor cells in 11 patients (nonresponders). In responders,FDG uptake decreased by 72% 6 11%; in nonresponders, itdecreased by only 42% 6 22%. At a threshold of 52% de-crease of FDG uptake compared with baseline, sensitivity todetect response was 100%, with a corresponding specificityof 55%. The positive and negative predictive values were 72%and 100%. Nonresponders to PET scanning had a signifi-cantly worse survival after resection than responders.

ConclusionFDG-PET is a valuable tool for the noninvasive assessment ofhistopathologic tumor response after neoadjuvant radiother-apy and chemotherapy.

Most patients with esophageal squamous cell carcinomahave locally advanced disease at first diagnosis. Surgery

remains the therapy of choice, provided a macroscopicallyand microscopically complete resection can be accom-plished (R0 resection).1,2 Preoperative chemotherapy andcombined radiotherapy and chemotherapy have been intro-duced with the primary objective of increasing the rate ofcomplete resections by downsizing the primary tumor, withimprovement of local tumor control and prevention of distantmetastases.3,4Nonresponders to neoadjuvant treatment seem tohave a worse prognosis and disease-free and overall survivalrates after complete resection compared with patients withcomparable tumor stages treated by surgery alone.5

Part of the preliminary results of this study were presented at the 23rdGerman Cancer Congress in 1999 (published as an abstract in theJournal of Cancer Research and Clinical Oncology, Supplement toVol. 124, page 26) and at the 116th Annual Meeting of the GermanSociety of Surgery in 1999.

Correspondence: Dr. Björn L.D.M. Brücher, Chirurgische Klinik und Po-liklinik, Klinikum rechts der Isar der TUM, Ismaninger Strasse 22,D-81675 München, Germany.

E-mail: [email protected]

ANNALS OF SURGERYVol. 233, No. 3, 300–309© 2001Lippincott Williams & Wilkins, Inc.

300

Morphologic imaging modalities (endoscopy, endoscopicultrasonography [EUS], computed tomography [CT], mag-netic resonance imaging) do not reliably differentiate be-tween viable tumor and inflammatory reactions, edema, andscar tissue.6–10 The need for noninvasive methods to dif-ferentiate between responders and nonresponders after neo-adjuvant therapy for esophageal carcinoma is widelyacknowledged.11

Positron emission tomography (PET) using [18F]-deoxy-glucose (FDG) has been used to visualize enhanced glucoseutilization in tumor tissue. Based on this biochemical signal,FDG-PET has been shown to identify primary tumors, re-gional lymph nodes, and distant metastases with consider-able diagnostic accuracy.12,13Markedly increased FDG up-take in squamous cell esophageal cancer has beendocumented in several studies.14–17 In breast cancer, se-quential FDG-PET imaging has been shown to provide asensitive means of early detection of tumor response totherapy.18,19 However, no information is available describ-ing the role of FDG-PET for the noninvasive assessment ofresponse to neoadjuvant therapy in esophageal cancer.

The purpose of this study was to use FDG-PET imagingprospectively in patients with locally advanced esophagealsquamous cell cancer to assess the response to preoperativeneoadjuvant radiotherapy and chemotherapy. Endpointswere the comparison of reduction in tumor FDG uptakewith postoperative histopathologic tumor response andsurvival.

METHODS

Between November 1995 and March 2000, 27 patientswith histologically proven esophageal squamous cell carci-noma were included in this study. Exclusion criteria werediabetes, peripheral vascular and coronary artery disease, orother contraindications for esophagectomy.20,21Patients re-ceiving prior treatment (chemotherapy, radiotherapy, lasertherapy, or stent implantation) were not included. To deter-mine the tumor stage precisely, patients underwent endos-copy, EUS, bronchoscopy (including brush cytology andbiopsy), and CT. Patients with locally advanced esophageal

carcinomas without distant metastases (uT3/4, N0/1, M0),located at or above the tracheal bifurcation, who wereeligible for surgery were included, and a baseline FDG-PETwas obtained.

After these diagnostic procedures, patients underwentsimultaneous radiotherapy and chemotherapy for 3 weeks.Assessment of tumor response by CT, endoscopy, EUS, andfollow-up PET scan was performed 3 weeks after comple-tion of therapy. Patients without signs of tumor progressionor other contraindications for esophagectomy were sched-uled to undergo surgery. Figure 1 summarizes the timeschedule of the study protocol. The study protocol wasdesigned to allow comparison of FDG-PET data and his-topathologic evaluation of tumor tissue at close time points.It was approved by the institutional review board at theTechnische Universität München. Written informed consentwas obtained from all patients.

Neoadjuvant Treatment

Neoadjuvant therapy consisted of simultaneous radiother-apy and chemotherapy. The external radiotherapy was de-livered by a two-field technique using a 10- to 15-MVphoton beam of 2 Gray per fraction (per day), 5 fractions perweek, to a total dose of 30 Gray. Simultaneously, chemo-therapy with 5-fluoruracil was administered as a continuousinfusion for 21 days at a daily dose of 300 mg/m2.22 Thisregimen was chosen because in our experience it minimizesside effects while still producing response rates comparableto more aggressive schedules.1,3,23,24

Clinical Response Evaluation

The clinical response evaluation was performed 3 weeksafter the end of therapy based on the criteria defined by theWorld Health Organization.25 The assessment of clinicalresponse included repeated endoscopy, EUS, and CT scan.Endoscopic and EUS response was assessed by one inves-tigator, judging the response based on macroscopic aspectsof the tumor mass. The findings of the endoscopic investi-

Figure 1. Time schedule of the study protocol.

Vol. 233 ● No. 3 Brücher and Others 301

gations were documented on video. CT scans were obtainedusing a conventional technique after oral and intravenousadministration of contrast. Pharmaceutical hypotonia wasachieved by giving 40 mg butylscopolamine before scan-ning. Response was assessed by one experienced radiolo-gist, measuring maximal tumor length and maximum wallthickness before and after treatment. Results of endoscopy,EUS, CT scan, and the clinical workup were discussed bythe investigators, and response was defined by consensus.The results of the PET study were not known to the clinicalinvestigators at the time of clinical response evaluation. Ifpatients were found to have inoperable disease after neoad-juvant treatment, further palliative treatment was given.

PET Imaging

Patients fasted overnight before PET scanning to mini-mize the blood insulin level and to ensure standardizedglucose metabolism in all patients. Blood glucose levelswere measured before each PET examination. FDG withhigh specific activity was synthesized by nucleophilic flu-orination.26 For each PET scan, patients received an intra-venous injection of 250 to 370 MBq FDG. Imaging wasperformed on an ECAT 951/R or ECAT EXACT PETscanner. (Siemens CTI, Knoxville, TN). Emission images ofthe tumor region of 20 minutes’ duration were acquired 40minutes after tracer injection. A transmission scan for at-tenuation correction was acquired after the emission scan.The duration of the transmission scan was 10 to 15 minutes,and approximately 4 million counts were acquired per slice.Emission data corrected for random events, dead time, andattenuation were reconstructed with filtered back-projection(Hanning filter; cutoff frequency 0.4 cycles/pixel, slicethickness 3.4 mm). The image counts/pixel were calibratedto activity concentration (Bq/mL). The resulting spatialresolution was approximately 8 mm at full width at halfmaximum.

Data Analysis

Attenuation-corrected PET images were normalized forthe injected dose of FDG and the patient’s body weight,resulting in parametric images representing regional stan-dardized uptake values (SUVs)27: SUV 5 (tissue activityconcentration)3 (body weight)/(administered activity). Forthe quantitative evaluation of regional FDG uptake, regionsof interest (ROIs) were manually placed over all tumors. Acircular ROI of 1.5 cm was placed in the slice with maxi-mum FDG uptake. If no focal FDG uptake was visible in thefollow-up examinations, the ROI was placed in the samelocation as the previously identified lesion using the land-marks of the transmission images (apex of the lungs, bifur-cation of the trachea) as a reference. SUVs were calculatedusing the average activity values in the ROI.

Surgical Therapy

All patients who were considered to be suitable for sur-gery and did not demonstrate tumor progression after neo-adjuvant therapy were referred to surgery. Surgical therapyconsisted of transthoracic en bloc esophagectomy with two-field lymphadenectomy in tumors located at and above thetracheal bifurcation.28,29 Cervical tumors were treated witha partial esophageal resection (sleeve resection).30 Gastro-intestinal continuity was maintained by a gastric pull-up inpatients with tumors located at and above the trachealbifurcation and with a free jejunal graft in patients withcervical tumors.

Histopathologic Evaluation

Resection specimens were fixed with formaldehyde (4%)for 24 hours. The complete tumor was cut into slices con-taining the entire esophagus wall marked with ink at allresection margins. Lymph nodes were prepared from theremaining perimuscular tissue, and the oral and aboral re-section margins were sampled. The tissue was paraffin-embedded and serial sections of each block were cut (5mm)and stained with hematoxylin and eosin and van Giesonstain. All specimens were classified by two experiencedpathologists unaware of the clinical and PET data in accor-dance with the criteria of the International Union AgainstCancer,31 including the R classification and grading. Theresponse to treatment was classified as complete, subtotal,partial, minimal remission, and no change in accordancewith the criteria described by Mandard et al,32 with modi-fications. Briefly, the percentage of viable residual tumorcells was assessed as follows. Complete response showedhistologic fibrosis with or without inflammation extendingthrough the different layers of the esophageal wall, but withno viable residual tumor cells. Subtotal response was char-acterized by the presence of less than 10% viable residualtumor cells, partial response by 10% to 50%, and minimalremission by more than 50% of viable residual tumor cells.No change was defined by the absence of any regressivechanges. For statistical analysis, the histopathologic re-sponse of patients undergoing resection was classified intotwo groups: group 1 (responder) consisted of patients withcomplete and subtotal responses, and group 2 (nonre-sponder) included patients with partial and minimal re-sponse and no change.

Statistical Analysis

PET signals were expressed as mean6 standard devia-tion, and differences between groups were tested with vari-ance analysis ort test. For intraindividual comparison oftumor FDG uptake parameters before and after neoadjuvanttherapy, a pairedt test was used. Survival rates were cal-culated according to Kaplan-Meier33 and tested with a log-

302 Response Monitoring in Neoadjuvant-Treated Patients With Esophageal Cancer by PET Ann. Surg. ● March 2001

rank test.34 Differences in proportion of patients were ana-lyzed by a chi-square Test. All tests were two-sided.P ,.05 was considered significant.

To evaluate the diagnostic accuracy of a reduction inFDG uptake for assessment of histopathologic response, areceiver operating characteristic curve was calculated. Thiscurve shows, for all observed changes of tumor FDG up-take, the percentage of responding and nonresponding tu-mors with an equal or greater reduction of FDG uptake.Thus, the sensitivity and specificity of a reduction in FDGuptake for assessment of histopathologic response can bedetermined for various cutoff values. In the present study,sensitivity was defined as the proportion of respondingtumors that showed a reduction in FDG uptake that wasequal to or greater than the cutoff value. The correspondingspecificity was defined by the proportion of nonrespondingtumors that showed a smaller decrease in FDG uptake thanthe cutoff value. By this definition, the positive predictivevalue of FDG-PET is the percentage of tumors with agreater decrease in FDG uptake than the cutoff value thatprove to be responding. The negative predictive value is thepercentage of tumors with a smaller decrease in FDG uptakethan the cutoff value that prove to be nonresponding. Fi-nally, the accuracy of PET imaging was defined as thenumber of correctly identified responding and nonrespond-ing tumors divided by the total number of examined tumors.For all estimated parameters, 95% confidence intervals werecalculated.

RESULTS

Demographic Data

The median age of the 27 patients was 52.9 years6 6.1(range 37.8–61). There were 4 female and 23 male patients.The primary tumor category on pretherapeutic staging wasuT2 in 2 patients, uT3 in 23 patients, and uT4 in 2 patients.Seven (29.2%) tumors were located in the cervical esoph-agus and 17 (70.8%) in the intrathoracic esophagus. Eight(29.6%) patients had histologically well-differentiated tu-mors (G2) and 19 (70.4%) had moderately differentiatedtumors (G3).

Clinical Response Evaluation

One patient (3.7%) had a complete remission, 11 patients(40.7%) had a partial remission, 5 patients (18.5%) had aminor remission, and 10 patients (37.1%) showed nochange. Twenty-four patients (88.9%) were referred forsurgery. All patients with complete, partial, and minor re-sponse and seven patients with no change underwent sur-gery. Three patients (11.1%) with no change received fur-ther palliative treatment because of new alcohol abuse andfunctional inoperability (Fig. 2). For further analysis, pa-tients with complete and partial remissions (n5 12) were

classified as clinical responders. All other patients wereconsidered nonresponders (n5 15). No influence of tumorlocation on clinical response was observed (P 5 .18)

Surgery

All 24 patients referred to surgery after preoperativetreatment underwent resection. A complete resection (R0)was achieved in 20 patients (83.4%), whereas 4 patients(16.6%) had histopathologically residual tumor in the radialresection margin (tumor bed). In seven patients (29.6%)with cervical carcinoma, an interposition of a free jejunalgraft was performed. Therefore, 17 patients (70.4%) under-went a transthoracic en bloc resection and two-field lymph-adenectomy followed by a reconstruction with a gastric tubewith a left cervical anastomosis. The postoperative compli-cation rate was 37.5% (9/24). One patient died after surgeryof a myocardial infarction.

Histopathologic Workup and ResponseEvaluation

The distribution of the pT, pN, and pM category and thestage grouping is shown in Table 1. The histopathologicresponse evaluation revealed complete response in 3 of 24patients (12.5%), subtotal response in 10 patients (41.7%),partial response in 6 patients (25%), and minimal responsein 5 patients (20.8%). There were no patients who had nochange. Thus, group 1 (complete and subtotal response)consisted of 13 patients, group 2 (partial and minimal re-sponse) of 11 patients. Sixteen patients (66.7%) had nolymph node metastasis (pN0), whereas eight patients(33.3%) had distant lymph node metastasis. Comparing thepretherapeutic clinical staging with the postoperative histol-ogy, the neoadjuvant therapy was associated with down-

Figure 2. Clinical response to neoadjuvant therapy.


staging of the primary tumor (T category) in 45.8%. Noinfluence of tumor location on histopathologic response(P 5 .48) was observed. The sensitivity and specificity ofclinical response for assessment of histopathologic responsewere 55% and 54%, respectively.

FDG Uptake Before and After Treatment

All 27 patients underwent imaging before and after treat-ment, providing a total of 54 PET scans. There was nosignificant difference in plasma glucose levels at the time oftracer injection before treatment (966 15 mg/100 mL) andafter treatment (946 13 mg/100 mL). All primary tumorsdemonstrated intensive FDG uptake at baseline. The SUVaveraged 8.36 2.5 before treatment and decreased signif-icantly to 3.36 1.5 after treatment (P , .0001). Figure 3shows typical examples for FDG-PET studies obtained fromtwo patients before and after neoadjuvant therapy. Cervicaltumors had a mean decrease in FDG uptake of 47.6%6

14.8, whereas intrathoracic tumors showed a decrease of62.5%6 24.2 (P 5 .024).

Correlation of FDG-PET and ClinicalResponse

The decrease in SUV between both PET studies in pa-tients with varying degrees of clinical response (completeresponse, n5 1; partial response, n5 11; minimal re-sponse, n5 4; on change, n5 8) was significantly corre-lated (P 5 .006) with a decrease in SUV in responders(complete and partial responses, n5 12) of 68%6 9.7%versus 49.07%6 26.2% in nonresponders (minimal re-sponse and no change, n5 12).

Correlation of FDG-PET andHistopathology

There was a significant correlation between the decreasein FDG uptake and the individual histopathologic response

Table 1. POSTOPERATIVE PATIENTS CHARACTERISTICS

Patient ypT* ypN† ypM†† UICC§ RClinical

Response¶Histopathologic

Response**FirstSUV

SecondSUV

SUV Decrease(%)

1 2 0 0 IIA R1 no change subtotal 10 2.5 752 0 0 0 0 R0 partial complete 7.5 1.9 753 3 1 0 III R0 partial minimal 4.6 1.8 624 3 0 0 IIA R0 minor subtotal 12.9 2.1 845 3 1 0 III R1 no change minimal 8.4 6.8 196 3 1 0 III R1 no change subtotal 8.1 3.9 527 — — — — — no change — 7.7 5.7 268 3 1 0 III R0 minor partial 6.8 6.9 119 3 0 0 IIA R0 partial partial 5.7 2.8 5210 3 1 1 IV R0 no change minimal 8.2 5.1 3711 4 1 1 IV R0 no change minimal 9.4 4.7 5112 — — — — — no change — 9 3.7 5913 3 0 1 IV R0 no change minimal 6.9 4.1 4014 0 0 0 0 R0 partial complete 6.3 1.8 7215 0 1 0 IIB R0 partial subtotal 6.1 2.3 6316 2 0 0 IIA R0 partial partial 7.1 2.3 6817 0 0 0 R0 complete complete 10.9 1.8 8318 3 0 0 IIA R0 partial partial 9.1 4 5619 — — — — — no change — 11.6 4.2 6420 3 0 0 IIA R0 minor partial 3.2 2.7 1621 3 0 0 IIA R0 partial partial 11.4 4.1 6422 2 0 0 IIA R1 minor subtotal 4.2 2 5223 3 0 0 IIA R0 minor subtotal 9.8 2.2 7724 2 0 0 IIA R0 partial subtotal 13.3 2.2 8425 1 1 0 IIB R0 no change subtotal 9.8 1.5 8526 1 0 0 I R0 partial subtotal 6.9 2.3 6727 1 0 0 I R0 partial subtotal 10.2 3.8 63

SUV, standardized uptake value. Patients 7, 12, and 19 did not undergo surgery.* Postoperative T stage according to the UICC.31

† Postoperative N stage according to the UICC.31

†† Postoperative M stage according to the UICC.31 In cases of M1, distant lymph node metastasis only.§ Postoperative stages according to the UICC.31

R Postoperative residual tumor according to the UICC.31

¶ Clinical response according to the WHO criteria.25

** Histopathologic response according to modified criteria of Mandard et al.32


categories (P 5 .01). Specifically, patients with completeresponse (n5 3) and subtotal response (n5 10) (respond-ers) had a decrease in SUV of 72.2%6 11.3%, whereasthose with partial response (n5 6) or minimal response(n 5 5) (nonresponders) showed a decrease of only 42.4%6 22% (P 5 .002). There was little difference betweenpatients with complete response (277%) and subtotal re-sponse (271%), as well as between those with partial andminimal response (243%,242%). The individual data areshown in Table 1 and Figure 4. Measurements of the base-

line PET showed FDG uptake of 8.96 2.6 with a decreaseto follow-up PET of 2.36 0.7 in responders, and 7.86 2.3with a decrease to 4.26 1.5 in nonresponders.

Receiver operating characteristic curve analysis for as-sessment of histopathologic response by PET revealed anarea under the curve of 9,161 (P 5 .001 for comparisonwith an area under the curve of 5,000) for all patients whounderwent resection (Fig. 5). The area under the curve forpatients who underwent transthoracic resection was 9,714(P 5 .001). Demanding a sensitivity of 100% (no his-

Figure 3. Examples of FDG-PET studies of responding and nonresponding tumors. In the baseline study,both tumors show intense FDG uptake. In the responding patient, FDG uptake decreased to backgroundactivity after neoadjuvant therapy (complete response). In contrast, FDG uptake by the nonrespondingtumor shows little change after therapy (partial response).

Figure 4. Individual standardized uptake value data of responders and nonresponders, including meanstandardized uptake value 6 standard deviation.


topathologic responder is incorrectly classified as a Table2). The 95% confidence interval for specificity was 23%to 83%; for sensitivity it was 75% to 100%. Calculatingthe highest sensitivity and specificity (i.e., the data pointof the receiver operating characteristic curve with theminimum distance from the 0% false-positive rate andthe 100% true-positive rate) revealed a sensitivity of 85%and a specificity of 82%.

Survival

The median follow-up time was 3.7 years6 1.2 (range0.8–5) (11 of 27 patients were alive as of this writing). The

overall 1-, 2-, and 3-year survival rates were 63.8%6 9.7%,29.9% 6 10.6%, and 19.9%6 10.8%, with a mediansurvival of 18.7 months6 2 (confidence interval 14.8–22.6) (see Fig. 5). Tumor location had no significant influ-ence on survival (P 5 .051). Patients with a decrease in theFDG uptake of less than 52% had a significantly (P ,.0001) shorter median survival time (8.8 months6 2.7)compared with patients with an SUV decrease of 52%ormore (22.5 months6 2.5). Analyzing only patients whounderwent resection (n5 24) revealed a median survival of22.5 months6 2.4 in patients with a decrease in FDGuptake of 52% or more and 6.7 months6 5 in those with adecrease in FDG uptake of less than 52% (P , .0001) (Fig.

Figure 5. Receiver operating characteristic analysis for assessment of histopathologic response withpositron emission tomography. Demanding a sensitivity of 100% resulted in a specificity of 55% (cutoff value52% reduction of FDG uptake). (A) All patients who underwent resection. (B) All patients who underwenttransthoracic resection.

Table 2. CROSSTABULATION OF HISTOPATHOLOGICAL RESPONSE AND PET-RESPONSE

Cross-Tabulation

Pathohistologic Response

TotalResponder Nonresponder

SUV $ 52%Count n 5 13 n 5 5 n 5 18% within variable SUV 72.2 (5PPV) 27.8 100% within variable pathology 100 (5sensitivity) 45.5 75

SUV , 52%Count — n 5 6 n 5 6% within variable SUV — 100 (5NPV) 100% within variable pathology — 54.5 (5specificity) 25

TotalCount n 5 13 n 5 11 24% within variable SUV 54.2 45.8 100% within variable pathology 100 100 100

SUV, standardized uptake valve; PPV, positive predictive valve; NPV, negative predictive valve.


6). Investigating only patients who underwent transthoracicresection revealed a median survival of 34 months6 12 inpatients with a decrease in the FDG signal of 52% or moreand 12 months6 8 in patients with a decrease of less than52% (P 5 .003) (Fig. 7).

DISCUSSION

In esophageal cancer, clinical response evaluation afterneoadjuvant treatment is limited by the lack of noninvasivetechniques that allow differentiation between respondersand nonresponders.6–10 The results of the present studyshow that metabolic signals measured by FDG-PET aresensitive and specific for identifying responders to neoad-juvant therapy.

The decrease in metabolic activity in tumor tissue wasmore closely related to histopathologic outcome than dataderived from the clinical evaluation of patients using otherimaging approaches. This difference is best explained bythe limitations of morphologic criteria to define the ex-tent of malignant cells. In addition, neoadjuvant therapymay lead to inflammatory reactions associated withedema, which may mask the loss of tumor tissue. How-ever, the number of patients studied in this populationwas too small to provide a statistically valid comparisonof the imaging modalities for the assessment of responseto neoadjuvant therapy.

It has been hypothesized that biochemical changes intumor tissue induced by therapy precede the change intumor size.19,35,36Therefore, imaging of tumor metabolismis expected to provide a sensitive means of detecting re-sponse to therapy. A recent study showed that FDG uptakein lymphomas decreases as early as 7 days after the start of

therapy.35 Our data clearly show that the decrease in tumor-FDG uptake correlates well with the results of histopatho-logic response evaluation. This confirms experimental datashowing a close correlation between FDG uptake and theextent of viable tumor cells.37

Previous studies from our laboratory evaluating the re-producibility of the FDG signal showed that the tumorglucose utilization rate is stable without therapeutic inter-ventions for the time period of the chosen study protocol.The interstudy variability of repeated FDG measurementswas less than 20%.38 The reported decrease in SUV valuesin our study population is clearly beyond the range ofinterstudy variability of FDG uptake and is therefore spe-cific for a therapy-induced effect.

The histopathologic evaluation of the whole tumor bedwith adjoining areas served as the gold standard for assess-ing response to neoadjuvant therapy. This approach mini-mizes potential errors in histopathologic response evalua-tion resulting from posttherapeutic tumor heterogeneity. Apreviously published scoring system that has been shown tobe of prognostic relevance was used for classifying post-therapeutic tumor regression in the resected specimen.32

This scoring system is based on the ratio of fibrosis versusviable tumor. However, the differentiation of therapy-in-duced fibrosis from preexisting tumor desmoplasia may bedifficult in individual cases. Histopathologic assessmentmay overestimate tumor regression after neoadjuvant treat-ment, especially in patients with minimal response. How-ever, in patients with complete response, it cannot be ex-cluded that single viable tumor cells have been overlooked,although the complete tumor bed has been examined his-topathologically. Therefore, we grouped patients with com-plete and subtotal responses as responders and those in theremaining categories as nonresponders.

On the basis of these results and our previous experiencein squamous cell carcinoma of the esophagus, we suggest

Figure 7. Survival of all patients who underwent resection (n 5 24) andpatients who underwent transthoracic resection (n 5 17) for groups ofpatients with different decreases in FDG uptake.

Figure 6. Overall survival of all patients (n 5 27).


the following protocol for the use of PET in the manage-ment of patients after neoadjuvant treatment. Respondersidentified by PET should undergo a consecutive resection.Nonresponders should not be referred to surgery. This issubstantiated by two observations. First, microscopic resid-ual tumor tissue cannot be detected by PET, suggesting thatan R0 resection is still required in responders. Second,nonresponders have such a poor prognosis (median survival9 months) that it appears questionable whether they wouldbenefit from surgical resection. However, this concept needsto be confirmed prospectively in larger patient groups todefine the role of PET imaging compared with conventionalresponse evaluation.

In conclusion, FDG-PET is a promising noninvasivetechnique for identifying response to neoadjuvant therapy inpatients with squamous cell carcinoma of the esophagus andmay significantly affect patient management.

Acknowledgments

The authors thank Raymonde Busch, MS, for help with the statisticalanalysis, Hans-Joachim Dittler, MD, for surgical endoscopy, and LeishaTyndale-Hines for editorial help in the preparation of this manuscript.They gratefully acknowledge the effort of the technologists and thecyclotron staff.

References

1. Fink U, Stein HJ, Siewert JR. [Multimodal therapy of tumors of theupper gastrointestinal tract]. Chirurgie 1998; 69:349–359.

2. Siewert JR, Holscher AH. Current strategy in surgery for esophagealcancer. Ann Ital Chir 1992; 63:13–18.

3. Fink U, Stein HJ, Bochtler H, et al. Neoadjuvant therapy for squamouscell esophageal carcinoma. Ann Oncol 1994; 5(suppl 3):17–26.

4. Raoul JL, Le Prise E, Meunier B, et al. Neoadjuvant chemotherapy andhyperfractionated radiotherapy with concurrent low-dose chemother-apy for squamous cell esophageal carcinoma. Int J Radiat Oncol BiolPhys 1998; 42:29–34.

5. Law S, Fok M, Chow S, et al. Preoperative chemotherapy versussurgical therapy alone for squamous cell carcinoma of the esophagus:a prospective randomized trial. J Thorac Cardiovasc Surg 1997; 114:210–217.

6. Zuccaro G Jr, Rice TW, Goldblum J, et al. Endoscopic ultrasoundcannot determine suitability for esophagectomy after aggressive che-moradiotherapy for esophageal cancer. Am J Gastroenterol 1999;94:906–912.

7. Dittler HJ, Fink U, Siewert GR. Response to chemotherapy in esoph-ageal cancer. Endoscopy 1994; 26:769–771.

8. Isenberg G, Chak A, Canto MI, et al. Endoscopic ultrasound inrestaging of esophageal cancer after neoadjuvant chemoradiation. Gas-trointest Endosc 1998; 48:158–163.

9. Jones DR, Parker LA Jr, Detterbeck FC, et al. Inadequacy of computedtomography in assessing patients with esophageal carcinoma afterinduction chemoradiotherapy. Cancer 1999; 85:1026–1032.

10. Laterza E, de Manzoni G, Guglielmi A, et al. Endoscopic ultra-sonography in the staging of esophageal carcinoma after preoper-

ative radiotherapy and chemotherapy. Ann Thorac Surg 1999;67:1466 –1469.

11. Kok TC. Chemotherapy in oesophageal cancer. Cancer Treat Rev1997; 23:65–85.

12. Weber WA, Avril N, Schwaiger M. Relevance of positron emissiontomography (PET) in oncology. Strahlenther Onkol 1999; 175:356–373.

13. Rigo P, Paulus P, Kaschten BJ, et al. Oncological applications ofpositron emission tomography with fluorine-18 fluorodeoxyglucose.Eur J Nucl Med 1996; 23:1641–1674.

14. Block MI, Patterson GA, Sundaresan RS, et al. Improvement instaging of esophageal cancer with the addition of positron emissiontomography. Ann Thorac Surg 1997; 64:770–777.

15. Flanagan FL, Dehdashti F, Siegel BA, et al. Staging of esophagealcancer with18F-fluorodeoxyglucose positron emission tomography.AJR Am J Roentgenol 1997; 168:417–424.

16. Kole AC, Plukker JT, Nieweg OE, et al. Positron emission tomographyfor staging of oesophageal and gastroesophageal malignancy. Br JCancer 1998; 78:521–527.

17. McAteer D, Wallis F, Couper G, et al. Evaluation of18F-FDG positronemission tomography in gastric and oesophageal carcinoma. Br JRadiol 1999; 72:525–529.

18. Schelling M, Avril N, Nahrig J, et al. Positron emission tomographyusing [18F]fluorodeoxyglucose for monitoring primary chemotherapyin breast cancer. J Clin Oncol 2000; 18:1689–1695.

19. Wahl RL, Zasadny K, Helvie M, et al. Metabolic monitoring of breastcancer chemohormonotherapy using positron emission tomography:initial evaluation. J Clin Oncol 1993; 11:2101–2111.

20. Bartels H, Stein HJ, Siewert JR. Preoperative risk analysis and post-operative mortality of oesophagectomy for resectable oesophagealcancer. Br J Surg 1998; 85:840–844.

21. Bartels H, Stein HJ, Siewert JR. Risk analysis in esophageal surgery.Recent Results Cancer Res 2000; 155:89–96.

22. Ajani JA, Ryan B, Rich TA, et al. Prolonged chemotherapy forlocalised squamous carcinoma of the oesophagus. Eur J Cancer 1992:28A:880–884.

23. Fink U, Stein HJ, Wilke H, et al. Multimodal treatment for squamouscell esophageal cancer. World J Surg 1995; 19:198–204.

24. Siewert JR, Brucher BL, Stein HJ, et al. [Esophagus carcinoma–systemic or local risk of recurrence–which perioperative measures aresuccessful?]. Langenbecks Arch Chir Suppl Kongressbd 1998; 115:290–294.

25. Miller AB, Hoogstraten B, Staquet M, et al. Reporting results of cancertreatment. Cancer 1981; 47:207–214.

26. Hamacher K, Coenen HH, Stocklin G. Efficient stereospecific synthe-sis of no-carrier-added 2-[18F]-fluoro-2-deoxy-D-glucose using amin-opolyether supported nucleophilic substitution. J Nucl Med 1986;27:235–238.

27. Strauss LG, Conti PS. The applications of PET in clinical oncology.J Nucl Med 1991; 32:623–650.

28. Akiyama H. Surgery for carcinoma of the esophagus. Curr Probl Surg1980; 17:53–120.

29. Siewert JR, Hölscher AH, Roder JD, et al. En-bloc esophagectomy inesophageal cancer. Langenbecks Arch Chir 1998; 373:367–376.

30. Steinau HU, Biemer E, Bader M, et al. Reconstruction of the cervicalesophagus by microsurgical transfer of intestinal segment. Berlin:Springer; 1988.

31. Sobin LH, Wittekind C. [TNM classification of malignant tumors. 5thed.]. New York: Wiley-Liss; 1997.

32. Mandard AM, Dalibard F, Mandard JC, et al. Pathologic assessment oftumor regression after preoperative chemoradiotherapy of esophagealcarcinoma. Clinicopathologic correlations. Cancer 1994; 73:2680–2686.

33. Kaplan EL, Meier P. Nonparametric estimation from incomplete ob-servations. J Am Stat Assoc 1958; 75:457–487.


34. Peto R, Pike MC, Armitage P, et al. Design and analysis of randomizedclinical trials requiring prolonged observation of each patient. II.analysis and examples. Br J Cancer 1977; 35:1–39.

35. Romer W, Hanauske AR, Ziegler S, et al. Positron emission tomog-raphy in non-Hodgkin’s lymphoma: assessment of chemotherapy withfluorodeoxyglucose. Blood 1998; 91:4464–4471.

36. Jansson T, Westlin JE, Ahlstrom H, et al. Positron emission tomog-raphy studies in patients with locally advanced and/or metastatic breast

cancer: a method for early therapy evaluation? J Clin Oncol 1995;13:1470–1477.

37. Higashi T, Tamaki N, Torizuka T, et al. FDG uptake, GLUT-1 glucosetransporter and cellularity in human pancreatic tumors. J Nucl Med1998; 39:1727–1735.

38. Weber WA, Ziegler SI, Thodtmann R, et al. Reproducibility of met-abolic measurements in malignant tumors using FDG PET. J NuclMed 1999; 40:1771–1777.


neoadjuvant therapy of esophageal squamous cell carcinoma: response evaluation by positron emission...

Documents