Abstract
Standard treatment for patients with anal cancer is concurrent chemoradiotherapy (CRT), the radiotherapy is combined with Mitomycin C (MMC) and 5-Flourouracil (5-FU). This treatment is known to cause high rates of severe hematologic toxicity (HT), which is associated with infections, fatigue and bleeding, and may lead to longer treatment times and reduction in chemotherapy dose. Longer treatment times and reduced chemotherapy dose may lead to suboptimal treatment outcome. This thesis studied 84 patients with histologically verified anal squamous cell carcinoma. Weekly blood cell counts during treatment, a 18-F-fluorodeoxyglucose-positron emission tomography/computed tomography (FDG-PET/CT) scan acquired before treatment and 3D radiotherapy dose maps were available for all patients. The aim of the study was to test if irradiation of active bone marrow (ABM), as defined by FDG-PET, was the principal cause of HT. The necessary code for reading and analysing the data was developed in MatLab. Total bone marrow (TBM) was defined using a Hounsfield threshold on the CT images. Standardized uptake values (SUV) were quantified for all voxels included in TBM, and then normalized by comparison with liver SUV. ABM was tentatively defined as the subvolumes of TBM exhibiting the 75%, 50% and 25% highest FDG-uptake, named ABM75, ABM50 and ABM25 respectively. Median dose was found to all these subvolumes. The patients were then split into a training set (59 patients) and a validation set (25 patients). Regression analyses were performed on the training set using the lymphocytes nadir, i.e. the lowest measured lymphocytes count, as response variable. Linear regressions found that median dose to TBM, ABM75, ABM50 and ABM25 were all significantly correlated with the lymphocytes nadir (p<0.05). The lowest root-mean-square-error (RMSE) was found for ABM25 (0.146 10^9/L), and the highest for TBM (0.160 10^9/L). However, after bootstrapping with 1000 iterations to find 95% confidence intervals (CI) of RMSE for both TBM and ABM25, the model using ABM25 did not perform significantly better than the model using TBM. The fitted models were then tested on the validation set by using receiver operating characteristic (ROC) curves. The ROC curves showed that both the ABM25 model and the TBM model were able to predict HT better than a random model, but there was no significant difference between the ABM25 and TBM models. In conclusion, irradiation of pelvic bone marrow was associated with HT. However, models using ABM, as defined by FDG-PET, did not improve model performance compared to models using TBM. One reason for this might be the strong correlation between dose to ABM and dose to TBM. Future clinical trials, aiming to spare the ABM during radiotherapy of anal cancer, should be conducted to test whether ABM truly reflects hematogenous tissue.