Approximation of head and neck cancer volumes in contrast enhanced CT

Abstract

BackgroundTumor volume may serve as a predictor of response to radiochemotherapy (RCT) in head and neck squamous cell carcinoma (HNSCC). Computer assisted tumor volumetry requires time-consuming slice-by-slice manual or semi-automated segmentation. We questioned how accurately primary tumor and suspect cervical lymph node (LN) volumes can be approximated by the maximum tumor diameters in three dimensions.MethodsIn contrast-enhanced diagnostic CT scans of 74 patients with incident advanced HNSCC, manual slice-by-slice segmentation volumetry of primary tumor, total- and largest suspect cervical LN served as the reference method. In the same scans, maximum orthogonal diameters were measured using the distance measurement tool in standard visualization software in axial and coronal sections. From these diameters, approximate volumes were calculated using the cubic and ellipsoid formula. A second segmentation volumetry was performed in contrast enhanced radiotherapy-planning CT scans obtained prior to primary concurrent RCT 24 days (+/− 13 days) following the initial diagnostic CT scans. Intraclass correlation coefficients and Bland-Altman analyses were used to compare results.ResultsSlice-by-slice manual segmentation volumetry of primary and LN volumes revealed a lognormal distribution and ranged from 0 to 86 ml and 0 to 129 ml, respectively. Volume approximations in diagnostic CT scans with the ellipsoid formula resulted in an −8 % underestimation of tumor volumes (95 % CI −14 % to −1 %; p = 0.022) and an −18 % underestimation of suspect cervical LN volumes (95 % CI −25 % to −12 %; p = 0.001). Inter rater intraclass correlation for primaries was 0.95 (95 % CI +0.92 to +0.97; p = 0.001), and intra rater intraclass correlation was 0.99 (95 % CI +0.98 to +0.99; p = 0.001). The cubic formula resulted in pronounced overestimation of primary and LN volumes. Primary tumor volumes obtained by the second segmentation volumetry in radiotherapy-planning CT scans obtained on average 24 days following the initial volumetry resulted in larger primary tumor volumes (mean bias +28 %, 95 % CI +14 % to +41 %; p = 0.001). Tumor volume increase correlated with time between the diagnostic and planning CTs (r = 0.24, p = 0.05) and was approximately 1 % per day.DiscussionEllipsoid approximations of tumor and lymph node volumes in HNSCC using maximum orthogonal diameters underestimates volumes based on segmentation in multiple slices. Due to time difference and safety margins, segmented volumes in radiotherapy-planning CT scans tend to be larger than in diagnostic CT scans.ConclusionEllipsoid approximations of tumor and lymph node volumes in HNSCC are easily available from diagnostic CT scans. Volume estimates are applicable over a wide range of tumor and LN sizes and may be useful in clinical decision-making and oncologic research.

Prognostic value of tumor volume in patients with head and neck squamous cell carcinoma treated with primary surgery

“…In stages I and II of HNSCC, 3D‐conformal RT was performed as nonsurgical treatment. The gross tumor volume encompassed all visible primary and nodal diseases and was delineated on the planning CT scans at the Department of Radiation Oncology with the PROSOMA Workstation (Oncology System Limited, Shrewsbury, UK), as previously described . An 8 mm 3D margin was generated for defining the clinical target volume for gross disease, irradiated up to 70 Gy.…”

Tumor volume as an independent predictive factor of worse survival in patients with oral cavity squamous cell carcinoma

“…The measurement of tumor volumes by imaging methods is known to be useful for treatment decision‐making and has been tested successfully for nonsurgical treatment modalities for tumors of the oropharynx and larynx . The complex calculation of manual slice‐by‐slice segmentation volumetry did not show superiority over the simple and straightforward method of multiplying the 3 tumor dimensions …”

“…21,24,[37][38][39] The complex calculation of manual sliceby-slice segmentation volumetry did not show superiority over the simple and straightforward method of multiplying the 3 tumor dimensions. 10 Our results are based on a cohort of 123 patients with surgically treated advanced oral cavity SCC, with a high mortality rate of 60.9% in at least 2 years of follow-up. Among the patients who survived and remained disease free, the average tumor volume was significantly lower.…”

“…Although imperfect, the estimate is reliable and reflects the actual tumor volume with reasonable accuracy. 10 The use of tumor volume as a predictor of survival or treatment responses has been already reported for head and neck tumors. [11][12][13][14][15][16][17] It is particularly relevant for nonsurgical treatment modalities.…”

“…A simple and straightforward method, previously described as the ellipsoid theory, consists of the simple multiplication of the 3 major diameters of the tumor, either obtained by imaging methods or by macroscopic measures of the operated tumor specimens. Although imperfect, the estimate is reliable and reflects the actual tumor volume with reasonable accuracy …”