Assessment of tumor size and tumour response for nasopharyngeal carcinoma (NPC): Implication of different methods of assessment for an Irregular shaped tumor

Abstract

5582 Background: Tumor size is an important outcome for phase II clinical trials. In the past bi-dimensional measurements (BDM) have been commonly used. More recently uni-dimensional measurements (UDM) have been adopted in the response evaluation in solid tumors (RECIST). However, for irregular shaped tumors such as NPC there remains the question as to whether these simple methods reflect the tumor size. The objective of this study is to evaluate the relation of UDM and BDM to total tumor volume (TTV). Methods: Patients undergoing conventional axial magnetic resonance imaging (MRI) of the head and neck for the staging of previously untreated NPC were recruited for this retrospective study. TTV, the summation of areas from cross sectional MRI is used as the reference standard. The UDM was obtained by measuring the maximum diameter of the tumor and BDM was obtained by UDM x (greatest measurement perpendicular to UDM). A general linear model was used to assess if TTV is associated with UDM or BDM. Comparisons of the response criteria RECIST (UDM CR<30%PR<SD<-20%PD), WHO (BDM CR<50%PR<SD<-25%PD) and change in TTV (CR<65%PR<SD<-40%) were made. Kappa statistic and McNemar’s test were performed to assess the degree of agreement and discordance in response. Results: We have 44 patients, of which 28 also had post-treatment measurements. TTV is associated significantly with BDM (p=0.0002) but not with UDM. The change from pre- to post-treatment in TTV is associated significantly with the change in BDM (p<0.0001) but not with the change in UDM. Percent change in TTV is associated significantly with percent change in BDM (p=0.0011) but not with UDM. In terms of response based on TTV, there are 4 CR, 16 PR, 8 SD. However, based on UDM there are 4 CR, 12 PR, 12 SD, kappa of 0.42 (95% CI 0.11 - 0.72). There are 7 PR based on TTV misclassified as SD and 3 SD misclassified as PR (p=0.1714). For BDM (WHO criteria) there are 4 CR, 13 PR, 11 SD, kappa of 0.47 (95% CI 0.16 - 0.77). There are 6 PR based on TTV misclassified as SD, and 3 SD misclassified as PR (p=0.2525). Conclusions: BDM gives a good reflection of the size of NPC, but the association was not significant in UDM. The response using UDM and BDM are comparable. No significant financial relationships to disclose.