Using¹⁸F-FDG PET/CT To Estimate The Length Of Primary Tumor And The Short-term Outcome To Chemoradiation Of Esophageal Squamous Cell Carcinoma Patients

Objective: To investigate the optimal threshold of using the¹⁸F-Fluorodeoxyglucose（FDG） positron emission tomography （PET）/computed tomography （CT） to determine thegross tumor length in nonoperative esophageal squamous cell cancer （ESCC） and delineatethe metabolic tumor volume of primary tumor to evaluate whether the metabolicparameters predict short-term outcome of chemoradiotherapy （CRT） in ESCC. Methods:Twenty-four patients （male19, female5ï¼› age40-79year; stage â…¡A-â…£A） withhistologically confirmed ESCC were enrolled from August2010to December2011. Thelengths of primary tumor were estimate using nine different methods: standard uptakevalue （SUV）2.0,2.5,3.0,3.5, and35%,40%,45%,50%,55%of maximum standarduptake value （SUVmax） on FDG PET/CT imaging before treatment. The length of primarytumors on PET scanning were measured and recorded as L2.0, L2.5, L3.0, L3.5, L35, L40,L45, L50and L55, respectively, and compared with the standardized activity length ofgross tumor in vivo under electronic fiberendoscope （Lst） to evaluated the optimalthreshold SUV by paired-t test and pearson correlation analysis. Twenty two patients （male18, female4ï¼›median age62-year） were evaluated by¹⁸F-FDG PET/CT before andfollowing40Gy radiotherapy with a synchronic cisplatin-based chemotherapy. Themetabolic tumor volume （MTV）, SUVmax, SUVmean and metabolic tumor length （Lm） ofprimary tumor were measured by two experienced and independent nuclear medicinephysicians, both with over five years of experience in PET-CT imaging to predict theshort-term outcome evaluated according to Response Evaluation Criteria in Solid Tumor（RECIST）. Patients with an outcome of complete response （CR） or partial response （RP）were defined as responders, and those with an outcome of stable disease （SD） orprogressive disease （PD） were classified as nonresponders. We observed the relationshipsbetween the short-term outcome of concurrent chemoradiotherapy and the metabolicparameters, also other clinical characteristics, such as sex, age, tumor location, clinicalstage of the nonoperative esophageal squamous cell cancer. Results The mean （±SD） Lst was5.27±2.45cm. The length of primary tumor, being from short to long, were L2.0, L2.5,Lst, L3.0, L3.5, L35, L40, L45, L50and L55. Compared with Lst, there was no significantdifference only in L2.5and L3.0（Pï¹¥0.05）. All these groups were significantly correlatedwith Lst and the correlation coefficients were between L3.5, L2.5, L3.0and Lst （r=0.935,0.920,0.919, respectively, all P﹤0.01）. Analysis of subgroups, we found there was nosignificant difference between Lst and L2.5, L3.0, L35, with the maximum correlationcoefficients from L2.5（r=0.953） in the group of smaller SUVmax of primary tumor, whileno significant difference was found only between Lst and L3.0（Pï¹¥0.05, r=0.791） inlarger SUVmax group. The overall response rate was68.2%. Because the SUVmax ofprimary tumor before treatment was14.27±5.61, smaller than15, so we use SUV2.5asthe threshold to delineate the extent of ESCC, which was certified in our first study. Thedifference of SUVmax. SUVmean, MTV, and Lm before treatment between the respondersand nonresponders were not significant （all pï¹¥0.05）, while the significant difference weredemonstrated after initial CRT （all p﹤0.05）. Only the change of MTV and Lm ofpretreatment and intratreatment were significantly different between the responders andnonresponders （p﹤0.05）, with the cutoff of69.80%,30.96%, respectively, defined byROC curve analysis. The sensitivity and specificity of MTV and Lm change for predictingtumor response were73.3%,100%and46.7%,100%, respectively, with the area undercurve （AUC） of0.781,0.700, respectively. Other parameters failed to predict tumorresponse both in pre-treatment and intra-treatment, with the AUC less than0.5.Conclusions: For the smaller SUVmax of ESCC, SUV cutoff of2.5provided the closestestimation and the length of primary tumor measured by SUV3.0may be closer to theactivity length of gross tumor in vivo for larger SUVmax. MTV and metabolic tumorlength changes might be the predictor of treatment response in nonoperative ESCC. Thesedata needs further research to validate.