The predictive value of metabolic response to preoperative radiochemotherapy in locally advanced...

ORIGINAL ARTICLE

The predictive value of metabolic response to preoperativeradiochemotherapy in locally advanced rectal cancermeasured by PET/CT

Robert Rosenberg & Ken Herrmann & Ralf Gertler &

Beat Künzli & Markus Essler & Florian Lordick &

Karen Becker & Tibor Schuster & Hans Geinitz &

Matthias Maak & Markus Schwaiger &

Jörg-Rüdiger Siewert & Bernd Krause

Accepted: 12 November 2008 / Published online: 3 December 2008# Springer-Verlag 2008

AbstractBackground To evaluate the value of positron emissiontomography using fluorodeoxyglucose and computer to-mography scan (FDG-PET/CT) for prediction of histopath-ological response of preoperative radiochemotherapy(RCTX) in patients with rectal carcinoma.Methods Thirty patients with uT3 rectal carcinoma wereexamined by FDG-PET/CT at baseline, 14 days afterinitiation, and after completion of preoperative RCTX.

The FDG decreases seen with PET scanning from baselineto day 14 (early metabolic response) and after completionof therapy (late metabolic response) were compared withhistopathological tumor response. One patient deniedsurgery after RCTX.Results The mean (±SD) reduction of tumor FDG uptakein histopathologically responding compared to non-responding tumors was −44.3% (±20.1%) versus −29.6%(±13.1%) (p=0.085) at day 14 and −66.0% (±20.3%)versus −48.3% (±23.4%) (p=0.040) after completion ofRCTX. Best differentiation of histopathological tumorresponse was achieved by a cut-off value of 35% reductionof initial FDG uptake at day 14 and 57.5% after completionof therapy. Applying the cut-off values as a criterion formetabolic response, histopathological response was pre-dicted with a sensitivity of 74% (14/19) at day 14 and 79%(15/19) after completion of therapy. The positive predictivevalue for early metabolic response was 82% (14/17) and forlate metabolic response was 83% (15/18). Histopathologicalevidence of accumulated peritumoral inflammation cellswas associated with a minor FDG decrease in fivehistopathologically responding patients, and influenced theresults with negative predictive values of 58% (7/12) and64% (7/11) at the early and late time points, respectively.Conclusions Metabolic response to a preoperative RCTXusing FDG-PET/CT in rectal cancer patients can becorrelated with histopathological response, but FDG uptakeof peritumoral inflammation cells limited the results and ledto false negative results.

Keywords Rectal cancer . Preoperative radiochemotherapy .

PET. Response

Int J Colorectal Dis (2009) 24:191–200DOI 10.1007/s00384-008-0616-8

R. Rosenberg (*) : R. Gertler :B. Künzli : F. Lordick :M. Maak : J.-R. SiewertChirurgische Klinik und Poliklinik, Klinikum rechts der Isar derTechnischen Universität München,Ismaningerstr. 22,81675 Munich, Germanye-mail: [email protected]

K. Herrmann :M. Essler :M. Schwaiger :B. KrauseNuklearmedizinische Klinik und Poliklinik,Klinikum rechts der Isar, Technische Universität München,Munich, Germany

K. BeckerInstitut für Pathologie und Pathologische Anatomie,Klinikum rechts der Isar, Technische Universität München,Munich, Germany

T. SchusterInstitut für medizinische Statistik und Epidemiologie,Klinikum rechts der Isar, Technische Universität München,Munich, Germany

H. GeinitzKlinik für Strahlentherapie und Radiologische Onkologie,Klinikum rechts der Isar, Technische Universität München,Munich, Germany

Introduction

Preoperative radio(chemo)therapy in locally advancedrectal carcinoma is an established standard procedure sincethe publication of randomized multi-center trials [1–3].These studies showed a significantly reduced risk for localrecurrence after preoperative radio(chemo)therapy, butfailed to demonstrate a significant improvement in overallsurvival.

Bakx et al. were able to demonstrate that only a smallpercentage of patients benefit from a preoperative radio-therapy in patients with resectable rectal carcinomas [4].They classified the patients of four randomized clinicaltrials according to their benefits (reduction of localrecurrence) and harms (increase of short-term complica-tions). Most patients had neither benefit nor additionalharm, while only 5–15% of patients had a benefit withoutadditional harm. Therefore, the question raises if all patientswith locally advanced tumors should receive preoperativeradio(chemo)therapy.

Guillem et al., Rodel et al., and our group were able toshow the impact of histopathological tumor responseassessment after preoperative radiochemotherapy (RCTX)[5–7]. Subgroups of patients with histopathologicallycomplete tumor response and a significantly improvedoverall survival compared to patients with histopathologi-cally incomplete tumor response were identified.

Clinical consequences from histopathological responseevaluation can only be drawn postoperatively and mayresult in suggestions of an adjuvant therapy. However,individualized tumor therapy with respect to neoadjuvanttreatment is a future goal in the field of surgicaloncology. Therefore it would be desirable to predictresponse before or early in the course of preoperativeradio(chemo)therapy.

Response prediction before or early in the course ofpreoperative radio(chemo)therapy has been attempted byseveral groups with the use of molecular markers andmolecular imaging. Ghadimi et al., Watanabe et al., and ourown group have shown that response can be pretherapeuti-cally predicted by a gene expression profile using micro-arrays [8–10]. Larger trials are needed to validate theseresults. Our group has previously shown the impact of earlymetabolic response assessment by positron emission to-mography using fluorodeoxyglucose (FDG-PET) and itscorrelation with prognosis in patients with esophageal andgastric cancer [11–14]. Metabolic responders had a signif-icantly improved overall survival compared to non-responders [13]. Amthauer et al. and Kalff et al. correlatedFDG-PET before and after neoadjuvant RCTX withhistopathology in patients with rectal cancer [15, 16].Cascini et al. evaluated the use of FDG-PET 12 days afterstarting RCTX for early response evaluation [17]. This

group was the first being able to show that early FDG-PETresponse evaluation allows to predict pathologic responseto preoperative treatment in patients with rectal cancer.

We prospectively investigated in a pilot study patientswith cT3NxM0 rectal cancer who received a preoperativeRCTX. Our hypothesis was that responding carcinomasshow a decrease of glucose utilization early, i.e. 2 weeks,after initiation of therapy, while glucose utilization remainsunchanged in non-responding rectal carcinomas. Changesin glucose utilization, as measured by FDG-PET/computertomography (CT) before RCTX, 14 days after the start andafter completion of RCTX were compared with histopath-ological response in the resection specimens. We presentthe first study evaluating the changes in glucose utilizationusing FDG-PET in combination with a diagnostic spiralcomputer tomography, which measured FDG-uptake pre-operatively at three different time points in rectal cancerpatients.

Materials and methods

Patients

Our pilot study was performed prospectively betweenMarch 2006 and January 2007 evaluating the use of PETfor response prediction to preoperative RCTX in patientswith rectal cancer. Thirty patients agreed to participate inthe study during the time period and were included in thestudy. This study had been approved by our ethicscommittee. All patients gave their informed consent priorto their inclusion in the study. Eligibility criteria were thepresence of biopsy-proven rectal adenocarcinoma between0–15 cm from the anal verge with or without clinicalevidence of local lymph node metastases, a tumor staged asuT3Nx by endosonography, no distant metastasis (cM0), noprevious chemo- or radiotherapy, a Karnofsky performancestatus of more than 60%, normal liver and kidney function,and no signs of congestive heart failure. Exclusion criteriawere previous malignant tumor disease, lack of compliance,or contraindications to RCTX. Staging procedures includeda rectoscopy, endoscopic ultrasound, colonoscopy, magnet-ic resonance imaging of the pelvis, and whole-body FDG-PET/CT scan in every patient. All patients were classifiedas uT3NxM0. All staging examinations were presented atour interdisciplinary tumor conference to confirm theindication for preoperative RCTX.

Preoperative radiochemotherapy

Preoperative RCTX consisted of percutaneous radiotherapyof 45.0 Gy delivered in 25 fractions of 1.8 Gy, five timesweekly. Planning target volume consisted of the tumor area

192 Int J Colorectal Dis (2009) 24:191–200

with a craniocaudal margin of 5 cm each and a ventralmargin of at least 3 cm. A complete internal iliacal lymphaticdrainage was also included. A three- or four-field boxtechnique with conformal portals was used (6–15 MVphotons). Throughout the radiation period, chemotherapywith 5-FU was given as a continuous infusion in a doseof 250 mg/m2 per day on days 1–5 of every week over5 weeks [7].

Surgical therapy and response evaluation

Tumor staging as reported above was performed prether-apeutically and was repeated preoperatively 4 weeksafter preoperative RCTX for all patients. Surgery wasscheduled to be performed during week 5 after comple-tion of RCTX. All tumor resections included a nerve-preserving preparation with high ligation of the superiormesenteric artery, lymphadenectomy, and total mesorectalexcision (TME). Partial mesorectal excision (PME) wasperformed in proximal rectal cancer. The quality of PMEor TME (mercury criteria) was not assessed by ourpathologists.

Patients with a distal rectal carcinoma underwentabdominoperineal extirpation (13 cases) and low anteriorresection (two cases) depending on the presence ofsphincter infiltration or pelvic floor infiltration. All patientswith an anastomosis received a protective ileostomy.Fifteen patients with a rectal carcinoma in the middle orproximal third underwent anterior resection with primaryanastomosis.

All resection specimens were examined according to astandardized histopathological protocol that included theevaluation of the (y)pTNM categories according to theInternational Union against Cancer (UICC), the number ofresected and tumor-involved lymph nodes and tumorinfiltration of the oral, aboral, and circumferential resectionmargins. Histopathological tumor regression was classifiedaccording to the classification of Becker [7, 18]. The degreeof tumor regression in response to RCTX was based on thehistological determination of the percentage of residualtumor tissue in the microscopically identifiable tumor bed.Three regression grades were used according to Becker etal. [18]: Grade I=complete (0% residual tumor; Grade Ia)or subtotal tumor regression (<10% residual tumor pertumor bed; Grade Ib); Grade II=partial tumor regression(10–50% residual tumor), and Grade III=minimal or notumor regression (>50% residual tumor). Patients withhistopathological response grade I were defined as histo-pathological responders according to previous results,which showed a significant improved overall survival forthis patient subgroup [7]. Patients with histopathologicalresponse grade II and III were defined as histopathologicalnon-responders.

With magnetic resonance imaging (MRI) and endo-sonography, response was defined as more or less than 50%tumor volume reduction.

PET imaging

All patients underwent three FDG-PET/CT at differentpoints of time: pretherapeutically (baseline), 14 daysafter the onset of the preoperative radiochemotherapyand 4 weeks after completion of the preoperativeradiochemotherapy.

Patients underwent [18F] FDG PET/CT (fluorine-18flourodeoxyglucose) on a Sensation 16 Biograph PET/CTscanner (Siemens Medical Solutions). The CT acquisitionprotocol included a low-dose CT (26 mAs, 120 kV, 0.5 sper rotation, 5-mm slice thickness) from the base of theskull to mid thigh for attenuation correction with dilutedoral contrast (Telebrix 300 mg) followed by the PET scanand a diagnostic CT in portal venous phase 80 s afterinjection of i.v. contrast agent (Imeron 300) (160 mAs,120 kV, 0.5 s per rotation 5-mm reconstructed slicethickness). All patients received a rectal filling with anegative contrast agent (100–150 ml). All PET scans wereacquired in 3D mode with an acquisition time of 3 min perbed position. Images were reconstructed by an attenuation-weighted ordered-subsets expectation maximizationalgorithm (four iterations, eight subsets) followed by apost-reconstruction smoothing Gaussian filter (5-mm fullwidth at half maximum).

Patients fasted 6 h before the PET scan and bloodglucose levels were measured before each PET examina-tion. All measured values were less than 150 mg/dl andshowed no significant changes during RCTX. Staticemission images of the tumor region of 20 min durationwere acquired 40 min after intravenous injection of 300 to400 MBq of FDG. If PET scans showed sufficient contrastbetween tumor and surrounding tissues (SUVtumor wasmore than 1.35×SUVliver+2×standard deviation SUVliver)patients received a repeat PET scan 2 weeks after initiationof the RCTX and after the completion of the RCTX.Semiquantitative evaluation of the three PET studies wasdone by a circular region of interest (diameter 1.5 cm) withthe TrueD software (Siemens Medical Solutions) andnormalized for injected dose and patients BSA. In thefollow-up PET studies, the regions of interest were placedat the same position as in the baseline scan using theanatomical landmarks as a reference. For assessment ofreproducibility of the quantitative evaluation of FDG-PETstudies, all PET scans were analyzed by two independentobservers. For additional analysis the mean value of the twomeasurements was used.

Mean and maximum standardized uptake values (SUV)were calculated for each scan by using circular ROIs

Int J Colorectal Dis (2009) 24:191–200 193

(diameter 1.5 cm) with the TrueD software. The percen-tages of SUV decrease from baseline to day 14 PET/CTscan (early metabolic response) and to the PET/CT scanafter completion of RCTX were calculated and comparedwith the histopathological tumor response.

Statistical analysis

Statistical analyses were performed using SPSS software(version 14.0; SPSS, Chicago, IL, USA). Quantitativevalues were expressed as mean±standard deviation and

median. Comparisons of related metric measurements wereperformed using Wilcoxon-signed rank test and Mann–Whitney U test in case of two independent samples.Fisher’s exact tests were used for comparison of frequen-cies between patient groups. Receiver operating character-istic (ROC) analyses were performed to evaluate the mostreliable cut-off value in terms of correctly predictinghistopathological response by FDG uptake reduction. Exact95% confidence intervals were reported for sensitivity andspecificity of calculated cut-off values. All analyses wereperformed two-sided at a 5% level of significance.

Table 1 Correlation of clinicaland histopathological parame-ters with PET response

Number Percent Early PET Late PET

Response Non-response Response Non-response

SexM 20 67 13 7 12 8W 10 33 5 5 7 3Tumor locationProximal third 6 20 3 3 5 1Middle third 9 30 6 3 7 2Distal third 15 50 9 6 7 8OperationAnterior resection 16 53 10 6 13 3Miles procedure 13 43 7 5 5 8No resection 1 4 1 0 1 0UICC stage0 7 23 6 1 6 1I 6 20 3 3 4 2II 6 20 3 3 3 3III 9 30 5 4 5 4IV 1 3 0 1 0 1No resection 1 3 1 0 1 0ypT-category0 7 23 6 1 6 11 1 3 1 0 1 02 5 17 2 3 3 23 15 50 8 7 8 74 1 3 0 1 0 1No resection 1 3 1 0 1 0ypN-category0 19 63 12 7 13 61 7 23 4 3 4 32 3 10 1 2 1 2No resection 1 3 1 0 1 0cM-categoryM0 28 93 17 11 18 10M1lymph 1 3 0 1 0 1No resection 1 3 1 0 1 0Resection marginsR0 25 83 15 10 17 8Rx or 1 3 10 2 1 1 2R2 1 3 0 1 0 1No resection 1 3 1 0 1 0


Results

Thirty patients were included in this study (10 female, 20male; age 61±10 years). The primary tumor was located inthe proximal third of the rectum in six patients (20%), nine(30%) tumors were localized in the middle third and 15(50%) tumors were localized in the distal third of therectum (Table 1).

Twenty-nine (97%) of the patients underwent tumorresection, one patient refused tumor resection after com-pletion of preoperative radiochemotherapy. The tumorresection was histopathologically complete (R0) in 25patients, microscopically incomplete (R1) at the circumfer-ential resection margin in three cases and macroscopicallyincomplete (R2) due to para-aortic lymph node metastasesin one patient. The tumor stages were ypT0 in seven, ypT1in one, ypT2 in five, ypT3 in 15, and ypT4 in one patient.Nineteen patients showed no lymph node metastases(ypN0), whereas lymph node metastases were present in10 patients (seven ypN1 and three ypN2). A histopatho-logical tumor response (grade I) was confirmed in 19 of 29(66%) patients while 10 of 29 (34%) were classified ashistopathological non-responders.

No patient showed insufficient image contrast in theinitial PET scan. Therefore, no patient had to be excludedfrom the metabolic response evaluation. In the 30 patientsassessable for early metabolic response, mean FDG-uptakewas 9.5±3.8 SUV. In the second PET scan 14 days after theonset of preoperative RCTX, mean SUV uptake decreasedby 38.5%±21.7% to 5.5±1.9 SUV (p<0.001). In the thirdPET scan after completion of preoperative RCTX meanSUV uptake decreased in total by 59.2%±24.0% to 3.5±1.8SUV (p<0.001).

There was no statistically significant difference in baselineFGD-uptake between histopathological responders and non-responders (responders, mean FDG-uptake of 9.7±4.4 SUV;non-responders, SUV mean 8.8±2.0 SUV) (p=0.701). Thereduction of tumor FDG uptake (mean±SD) after 14 days oftherapy showed a trend to be different between histopatho-logically responding (−44.3%±20.1%) and non-respondingtumors (−29.6%±13.1%) (p=0.085), although not reachingstatistical significance. Mean SUV uptake was 5.0±1.9 inresponders and 6.2±1.9 in non-responders (p=0.151). Aftercompletion of therapy, the reduction of FDG uptake was−66.0%±20.3% in histopathologically responding tumorsand −48.3%±23.4% in non-responding tumors (p=0.040).Mean SUV uptake was 2.8±1.3 in responders and 4.5±2.1in non-responders (p=0.027).

ROC analysis

In the ROC analysis, a more than 35% reduction of FDGuptake 14 days after the onset of preoperative RCTX

(second PET scan) was found to provide the highestaccuracy for differentiation of histopathological respondingand non-responding tumors. The PET analysis aftercompletion of preoperative RCTX (third PET scan)revealed a cutoff value of 57.5% of the initial FDG uptakefor the separation of metabolic responders and non-responders (Fig. 1).

Response evaluation

Eighteen of 30 patients (60%) were classified as metabolicresponders in the second PET scan (early responseevaluation; cut-off value of 35%), whereas 12 of 30(40%) patients were classified as metabolic non-responders.An example is shown in Figs. 2 and 3. After completion ofpreoperative RCTX 19 of 30 patients (63%) were classifiedas metabolic responders in the third PET scan (preoperativeresponse evaluation; cut-off value of 57.5%) and 11 of 30patients (37%) were classified as metabolic non-responders.One of the metabolic responders refused operation aftercompletion of preoperative RCTX.

Correlation with early metabolic response

Of the 17 early metabolic responders, which underwentresection, 14 patients (82%) achieved a histopathologicalresponse in the resection specimen (Table 2). In contrast,only five of the 12 (42%) early metabolic non-respondersshowed a histopathological response. Of the 19 histopath-ological responders, 14 patients had an early metabolicresponse and seven of 10 histopathological non-responders

AUC SUVmean after 2 weeks = 0.700

AUC SUVmean before surgery = 0.737

Fig. 1 ROC curves for prediction of response by PET imaging. Greenline delta SUV mean baseline–FDG uptake 14 days after initiation ofradiochemotherapy; blue line delta SUV mean FDG uptake baseline–FDG uptake after completion of radiochemotherapy


had an early metabolic non-response. This resulted in asensitivity of 74% (95% CI, 0.48 to 0.91) and specificity of70% (95% CI, 0.34 to 0.94), respectively. Positivepredictive value for the early metabolic response was 82%(95% CI, 0.56–0.97). The overall accuracy was 72% (95%CI, 0.53 to 0.87).

Correlation with late metabolic response

Of the 18 preoperative metabolic responders, whichunderwent resection, 15 patients (83%) achieved a histo-pathological response in the resection specimen (Table 3).In contrast, only four of the 11 (36%) preoperativemetabolic non-responders showed a histopathological re-sponse. Of the 19 histopathological responders, 15 patientshad a preoperative metabolic response and seven of 10histopathological non-responders had a preoperative meta-bolic non-response. This resulted in a sensitivity andspecificity of 79% (95% CI, 0.54 to 0.94) and 70% (95%CI, 0.34 to 0.94), respectively. Positive predictive value forthe early metabolic response was 83% (95% CI, 0.58 to0.97). The overall accuracy was 76% (95% CI, 0.56 to0.90).

Negative predictive value

Five of 12 patients (42%) and four of 11 patients (36%)with histopathological tumor response had minor metabolicSUV changes ranging from +20% to −25% and thereforehad to be classified as metabolic non-responders at the earlyand late metabolic response evaluation. Only one of the fivepatients revealed a metabolic response at late responseevaluation. These results led to a low negative predictivevalue of early and late metabolic response of 58% (95% CI,0.27 to 0.85) and 64% (95% CI, 0.30 to 0.90), respectively.In contrast to the other specimens, histopathologically weobserved marked peritumoral inflammation, with manyfoamy histiocytes being derived from the mononuclearphagocytic system. These features, most frequently associ-ated with neoadjuvant treatment, included foamy histio-cytes, vascular changes, marked fibrosis, and acellularmucus lakes [18]. The peritumoral presence of foamyhistiocytes after RCTX in areas where tumor cells under-went apoptosis was correlated with intensive glucosemetabolism and is our explanation for persisting highSUV levels 14 days after initiation and after completionof RCTX (Fig. 4).

SUVmean= 10,7 SUVmean= 3,1 SUVmean= 2,1

PET/CT before PET/CT after PET/CT after radiochemotherapy 14 days radiochemotherapy

Fig. 2 Example of FDG-PET/CT studies in a patient withmetabolic response. FDG uptakedecreases 14 days after initiationand after completion of radio-chemotherapy. First row showsthe FDG-PET scan, second rowdisplays the corresponding CT-scans and the third row repre-sents the fusion of FDG-PETand CT


Magnetic resonance imaging and endosonography

In addition, we assessed the response to radiochemotherapyby MRI and endosonography before and after radiochemo-therapy. Using endosonography, 12 of 19 patients withhistopathological response were classified as respondersand seven of 10 patients with histopathological non-response were classified as non-responders. Using MRI,11 of 19 patients with histopathological response wereclassified as responders and seven of 10 patients withhistopathological non-response were classified as non-responders. Thus MRI and endosonography were inferior

in predicting histopathological response compared to FDG-PET, but revealed equal results for the prediction ofhistopathological non-response.

Discussion

Our prospective pilot study demonstrates that metabolicresponse measurement using FDG-PET/CT in patients withrectal cancer can be correlatedwith histopathological response14 days after initiation (early metabolic response) and aftercompletion (late metabolic response) of a preoperative RCTX.

Table 2 Correlation of histopathological response with early meta-bolic response by FDG-PET (cut-off value 35%)

Number Percent PET response

Early response Early non-response

Number Percent Number Percent

Histopathological responseResponder 19 63 14 74 5 26Non-responder 10 34 3 30 7 70No operation 1 3 1 100 0 0N 30 100 18 60 12 40

Table 3 Correlation of histopathological response with preoperative(late) metabolic response by FDG-PET (cut-off value 57.5%)

Number Percent PET Response

Late response Late non-response

Number Percent Number Percent

Histopathological responseResponder 19 63 15 79 4 21Non-responder 10 34 3 30 7 70No operation 1 3 1 100 0 0

30 100 19 63 11 37

SUVmean= 12,1 SUVmean= 10,6 SUVmean= 5,2

PET/CT before PET/CT after PET/CT after radiochemotherap 14 da s radiochemotherap

Fig. 3 Example of FDG-PET/CT studies in a patient withmetabolic non-response. FDGuptake is almost unchanged14 days after initiation anddecreases only slightly aftercompletion of radiochemother-apy. First row shows theFDG-PET scan. The second rowdisplays the corresponding CT-scans and the third row repre-sents the fusion ofFDG-PET and CT


We present the first study which describes statistical cut-offvalues of decrease in FDG uptake (35% for early metabolicresponse and 57.5% for late metabolic response) whichdifferentiate between histopathological tumor response andnon-response. Sensitivity and specificity of the cut-off valueswere marginally worse compared to our results for patientswith gastric and esophageal cancer [11–14, 19].

The correlation between changes in FDG uptake aftercompletion of a preoperative RCTX and tumor response inrectal cancer was previously reported by several groups[20–22]. Kalff et al. examined the value of FDG-PET afterneoadjuvant RCTX in patients with locally advanced rectalcarcinoma [16]. The authors found that the FDG changescan be scored in complete, partial, stable, and progressivemetabolic response and can be correlated with overallsurvival. Amthauer et al. were able to demonstrate thatFDG-PET after neoadjuvant RCTX and combined regionalhyperthermia can predict histopathological response inrectal carcinoma [15]. They found FDG-PET to be superiorcompared to endorectal ultrasound for late responseassessment. Our late metabolic response evaluations (aftercompletion of preoperative RCTX) revealed a cut-off valueof 57.5% in FDG-uptake decrease for differentiation ofhistopathological responding and non-responding carcino-mas. This cut-off value resulted in a sensitivity of 79%,specificity of 70%, and accuracy of 76%. The stage atwhich testing is performed has potential clinical conse-quences. Late metabolic response evaluation can result onlyin changes of adjuvant therapy. On the other hand, potentialimplications of terminating a neoadjuvant therapy, if thepatient would benefit from preoperative RCTX, are shortertime until operation, no harm from radio- and chemother-apy, and the possibility of changing therapy.

Besides the study of Cascini et al., no other studyevaluated the value of early metabolic response evaluationof a preoperative radio(chemo)therapy in rectal cancer.Cascini et al. were the first performing early metabolicresponse evaluation using 18F-FDG PET for the predictionof histopathologic tumor response in 33 patients with

locally advanced rectal cancer. No correlation with latemetabolic response evaluations and no survival data werepresented [17]. They applied the same regimen of pelvicradiation (45 Gy) as in our study, but used a moreaggressive chemotherapy with additional oxaliplatin andraltitrexed compared to our study. Responders wereidentified with a sensitivity of 100%, specificity of 87%,and accuracy of 100% by a decrease of the mean SUV of≥52% 12 days after the onset of RCTX.

Our results in early metabolic response evaluationcorrelate with the results of late metabolic responseevaluation and with histopathological response, which wasnot described before. The accuracy of late metabolicresponse was only marginally superior (only one moreidentified histopathological responder). These results sup-port the impact of early metabolic response evaluation andmay theoretically allow modifications of the preoperativetreatment regimen in those that do not respond. Ouridentified cut-off value 14 days after radiochemotherapywas lower (cut-off value 35%) compared to Cascini et al.reporting on a 52% threshold. This might be explained bythe study population, the less aggressive chemotherapyprotocol used in our patient cohort, or other methodologicaldifferences as imaging protocol and data analysis.

We detected five patients with histopathological tumorresponse, which were classified as early metabolic non-responders. Only one of these five patients was a metabolicresponder at the late response evaluation. In these patientswe observed in the tumor specimen the large accumulationof inflammation cells in areas where tumor cells underwentapoptosis. This phenomenon correlated with an intensiveglucose metabolism and led to a low false negativepredictive value (Fig. 4). It is known from other studiesthat inflammation cells have an increased FDG uptake [22].In our study, these inflammation cells led to false negativeresults. FDG uptake after RCTX was caused in five casesby inflammation cells and by tumor cells. This phenome-non limited the value of metabolic response evaluation inour study. Our other metabolic response evaluation studies

Fig. 4 Macroscopic and microscopic examples for one of the fivepatients with histopathological tumor response (regression grade 1b)and metabolic non-response. A marked inflammation with many

foamy histiocytes being derived from the mononuclear phagocyticsystem can be identified histologically


of chemotherapy in Barrett’s carcinoma and of RCTX inesophageal carcinoma did not observe such accumulationsof inflammation cells after chemo- or radiochemotherapy[11, 13, 14].

In the past, several molecular, metabolic or radiologicalmethods were evaluated for their value of response prediction[23–26]. We did evaluate the use of FLT-PET ([18F]3′-deoxy-3′-fluorothymidine) for the prediction of responseto preoperative radiochemotherapy in patients with rectalcancer [27]. We were able to demonstrate a significantlydecrease of FLT uptake after initiation and after completionof radiochemotherapy, but the degree of change in FLTuptake did not correlate with histopathological tumorregression. In another study we studied the use of geneexpression profiles for response prediction in rectal cancer.Using microarray chips we compared the transcriptionprofiles of pretherapeutical rectal biopsies with the histo-pathological response [10]. A gene expression signature of42 genes was identified, which allowed a correct classifica-tion of responders in 71% of cases with a specificity of 86%.Denecke et al. compared CT, MRI, and FDG-PET beforeand after completion of a preoperative radiochemotherapyfor response prediction in rectal cancer. FDG-PET had a highsensitivity and was as expected superior compared to CT andMRI in predicting late metabolic response [28]. We alsofound that FDG-PET was superior for the detection ofresponders compared to MRI and endosonography. Nomethod can reliably predict histopathological response of apreoperative RCTX in rectal cancer patients at this time.

The presented study shows that early metabolic tumorresponse evaluation using PET/CT can be correlated withhistopathological tumor response and with late metabolicresponse evaluation in locally advanced rectal carcinoma,but the results are limited due to low negative predictivevalue. We found that a cut-off value of 35% for earlyresponse evaluation and 57.5% for late response evaluationdistinguishes between histopathological responders andnon-responders. We detected histologically peritumoralinflammation cells, which limited the predictive value ofour results. Nevertheless, the cut-off values should bevalidated in further, larger prospective studies.

References

1. Kapiteijn E, Marijnen CAM, Nagtegaal ID et al (2001) Preoper-ative radiotherapy combined with total mesorectal excision forresectable rectal cancer. N Engl J Med 345:638–646

2. Sauer R, Becker H, Hohenberger W et al (2004) Preoperativeversus postoperative chemoradiotherapy for rectal cancer. N EnglJ Med 351:1731–1740

3. Bosset JF, Collette L, Calais G et al (2006) Chemotherapy withpreoperative radiotherapy in rectal cancer. N Engl J Med355:1114–1123

4. Bakx R, Emous M, Legemate DA et al (2006) Harm and benefitof short-term pre-operative radiotherapy in patients with resect-able rectal carcinomas. Eur J Surg Oncol 32:520–526

5. Guillem JG, Chessin DB, Cohen AM et al (2005) Long-termoncologic outcome following preoperative combined modalitytherapy and total mesorectal excision of locally advanced rectalcancer. Ann Surg 241:829–836

6. Rodel C, Martus P, Papadoupolos T et al (2005) Prognosticsignificance of tumor regression after preoperative chemoradio-therapy for rectal cancer. J Clin Oncol 23:8688–8696

7. Rosenberg R, Nekarda H, Zimmermann F et al (2008) Histopath-ological response after neoadjuvant radiochemotherapy in rectalcarcinoma is associated with improved overall survival. J SurgOncol 97:8–13

8. Ghadimi BM, Grade M, Difilippantonio MJ et al (2005)Effectiveness of gene expression profiling for response predictionof rectal adenocarcinomas to preoperative chemoradiotherapy. JClin Oncol 23:1826–1838

9. Watanabe T, Komuro Y, Kiyomatsu T et al (2006) Prediction ofsensitivity of rectal cancer cells in response to preoperativeradiotherapy by DNA microarray analysis of gene expressionprofiles. Cancer Res 66:3370–3374

10. Rimkus C, Friederichs J, Boulesteix AL et al (2008) Microarray-based prediction of tumor response to neoadjuvant chemoradio-therapy of patients with locally advanced rectal cancer. ClinGastroenterology Hepatology 6:53–61

11. Weber WA, Ott K, Dittler HJ et al (2001) Prediction of response topreoperative chemotherapy in adenocarcinomas of the esophag-gastric junction by metabolic imaging. J Clin Oncol 19:3058–3065

12. Ott K, Fink U, Becker K et al (2003) Prediction of response topreoperative chemotherapy in gastric carcinoma by metabolicimaging: results of a prospective trial. J Clin Oncol 21:4604–4610

13. Ott K, Weber W, Lordick F et al (2006) Metabolic imagingpredicts response, survival, and recurrence in adenocarcinoma ofthe esophagogastric junction. J Clin Oncol 24:4692–4698

14. Brucher BL, Weber W, Bauer M et al (2001) Neoadjuvant therapyof esophageal squamous cell carcinoma: response evaluation bypositron emission tomography. Ann Surg 233:300–309

15. Amthauer H, Denecke T, Rau B et al (2004) Response predictionby FDG-PET after neoadjuvant radiochemotherapy and combinedregional hyperthermia of rectal cancer: correlation with endorectalultrasound and histopathology. Eur J Med Mol Imaging 31:811–819

16. Kalff V, Duong C, Drummond EG et al (2006) Findings on 18F-FDG PET scans after neoadjuvant chemoradiation providesprognostic stratification in patients with locally advanced rectalcarcinoma subsequently treated by radical surgery. J Nucl Med47:14–22

17. Cascini GL, Avallone A, Delrio P et al (2006) 18F-FDG PET is anearly predictor of pathologic tumor response to preoperativeradiochemotherapy in locally advanced rectal cancer. J Nucl Med47:1241–1248

18. Becker K, Mueller JD, Schumacher C et al (2003) Histomorphol-ogy and grading of regression in gastric carcinoma treated withneoadjuvant chemotherapy. Cancer 98:1521–1530

19. Lordick F, Ott K, Krause BJ et al (2007) Use of PET to assessearly metabolic response and guide treatment of locallyadvanced adenocarcinoma of the oesophagus and oesophago-gastric junction: a prospective feasibility study. Lancet Oncol8:797–805

20. Capirci C, Rampin L, Erba PA et al. FDG-PET/CT reliablypredicts response of locally advanced rectal cancer to neo-adjuvant chemo-radiation therapy. Eur J Nucl Med Mol Imaging.2007 May 15; [Epub ahead of print]


21. Capirci C, Rubello D, Chierichetti F et al (2006) Long-termprognostic value of 18F-FDG PET in patients with locallyadvanced rectal cancer previously treated with neoadjuvantradiochemotherapy. Am J Roentgenol 187:202–208

22. Kao PF, Chou YH, Lai CW (2008) Diffuse FDG uptake in acuteprostatitis. Clin Nucl Med 33:308–310

23. Oku S, Nakagawa K, Momose T et al (2002) FDG-PET afterradiotherapy is a good prognostic indicator of rectal cancer. AnnNucl Med 16:409–416

24. Saw RP, Morgan M, Koorey D et al (2003) p53, deleted incolorectal cancer gene, and thymidylate synthase as predictors ofhistopathologic response and survival in low, locally advancedrectal cancer treated with preoperative adjuvant therapy. DisColon Rectum 46:192–202

25. Llamas-Elvira JM, Rodrigues-Fenandez AR, Gutierrez-Sainz J etal (2007) Flourine-18 flourodeoxyglucose PET in the preoperative

staging of colorectal cancer. Eur J Nucl Med Mol Imaging34:859–866

26. Wieder H, Rosenberg R, Lordick F et al (2007) Magneticresonance imaging in patients with rectal cancer before neo-adjuvant radiochemotherapy for prediction of tumor-free circum-ferential resection margins and long-term survival. Radiology243:744–751

27. Wieder HA, Geinitz H, Rosenberg R et al (2007) PET imagingwith (18F) 3′-deoxy-3′-flourothymidine for prediction of responseto neoadjuvant treatment in patients with rectal cancer. Eur J NuclMed Mol Imaging 34:878–883

28. Denecke T, Rau B, Hoffmann KT et al (2005) Comparison of CT,MRI and FDG-PET in response prediction of patients with locallyadvanced rectal cancer after multimodal preoperative therapy: isthere a benefit in using functional imaging? Eur Radiol 15:1658–1666


The predictive value of metabolic response to preoperative radiochemotherapy in locally advanced...

Documents

Transcript of The predictive value of metabolic response to preoperative radiochemotherapy in locally advanced...