Abstract Book

ESTRO 37

S539

characteristics showed significant associations with PET- and CT- slopes for FEV1 (p=0.017, p=0.056), COPD status (p=0.007, p=0.083), and age (p=0.073, p=0.074). Further associations for PET-slope were tumor stage (p <0.001), N stage (p=0.004), and chemotherapy regimen(p=0.041).

plan, MRI pre ) and at half therapy (fr 9, for the ART plan, ). GTVs were contoured by a single clinician on ). Based on the Poisson-like tumor control probability formula, assuming the volume regression as proportional to the fraction of killed cells and neglecting the inter- patient variability of the kinetic of their removal, the parameter TCP early = -ln[(1 – (V half / V pre )) Vpre ] was suggested as a robust surrogate of early response. The discriminative power of TCP early in predicting the tumor pathological response was quantified by ROC curves (AUC, sensitivity, specificity, positive and negative predictive values (PPV and NPV)); three end-points were considered: pathological complete response (pCR); pCR or clinical complete response without surgery (cCR); limited response (residual vital cells (RVC) in the surgical specimen >10%). Results Complete data were available for 65/74 pts: pCR, cCR and RVC>10% were 20, 2 and 19. The discriminative power of TCP early was moderately high with AUC values equal to 0.79 (95% CI:0.67-0.89), 0.81 (0.69-0.89) and 0.75 (0.62-0.84) for pCR, pCR+cCR and RVC>10% respectively (p<0.0005). TCP early was highly sensitive for all considered end-points (85%, 85%, 90% respectively) with very high NPV (90-94%). In figure 1, the rates of pCR+cCR and of RVC>10% grouped by TCP early quartiles are plotted, showing the ability of TCP early in discriminating the response. In Figure 2, the logistic regression between TCP early and the rate of pCR+cCR (p<0.0001, OR: 0.90 (95%CI: 0.84-0.96), H&L test:0.71) is plotted together with the true rates, grouped by quartiles. MRI half both MRIs and GTV volumes were assessed (V pre , V half

Conclusion While both CT and PET metrics demonstrated high linearity with dose, PET-slope was not only a better indicator of RP in comparison to CT-slope, but also in relation to dose metrics. However, the PET-slope metric was further improved upon by adding the MLD to predict RP (F1 score=0.83). In addition, univariate analysis indicated several patient characteristics, which may be useful for future multivariate studies in identifying patients with higher dose sensitivity in both PET and CT. PO-0976 Rectal cancer radiochemotherapy: pathological response predicted by modeling early tumor regression C. Fiorino 1 , C. Gumina 2 , P. Passoni 2 , A. Palmisano 3 , S. Broggi 1 , G.M. Cattaneo 1 , A. Di Chiara 3 , M. Mori 1 , R. Raso 1 , N. Slim 2 , F. De Cobelli 3 , R. Calandrino 1 , N. Di Muzio 2 1 San Raffaele Scientific Institute, Medical Physics, Milano, Italy 2 San Raffaele Scientific Institute, Radiotherapy, Milano, Italy 3 San Raffaele Scientific Institute, Radiology, Milano, Italy Purpose or Objective Predictive models based on the individual response during neo-adjuvant radio-chemotherapy (RCT) of rectal cancer (Rca) have huge potential in treatment individualization; up to now, models based on the response during RCT were not reported. The aim of this study was to individually assess a radiobiological parameter (TCP early ) based on early tumor regression; then, the ability of TCP early in predicting the tumor pathological response was investigated. Material and Methods Seventy-four consecutive patients (pts) treated according to our protocol (41.4 Gy in 18 fractions, 2.3 Gy/fr) delivering an adaptive (ART) concomitant boost on the residual tumor (GTV) in the last 6 fractions (3 Gy/fr, mean dose to GTV: 45.6 Gy) were considered. Chemotherapy consisted of oxaliplatin 100 mg/m 2 on days -14, 0 (start of RT), and +14, and 5-fluorouracil 200 mg/m 2 /d from day -14 to the end of RT. High resolution T2-weighted MRI were taken before RT (for the initial

Conclusion TCP early

, individually assessing the early response from tumor regression was able to accurately predict the tumor pathological response in a relatively large group of pts treated within an ART protocol with RCT for Rca. The