Clinical investigation: head and neck
Tumor perfusion rate determined noninvasively by dynamic computed tomography predicts outcome in head-and-neck cancer after radiotherapy

https://doi.org/10.1016/S0360-3016(03)00764-8Get rights and content

Abstract

Purpose

To investigate the value of CT-determined tumor perfusion as a predictive factor of local and regional failure and cause-specific survival in head-and-neck cancer treated by radiotherapy.

Methods and materials

In 105 patients, the perfusion of a primary head-and-neck squamous cell carcinoma was estimated using dynamic CT. A contrast agent bolus was rapidly injected i.v., while during the first pass a dynamic data acquisition was performed at the level of the largest axial tumor surface. The perfusion in the selected tumor region of interest was calculated by dividing the slope of the tumor–time density curve by the maximal value in arterial density. Primary and nodal tumor volume was calculated from the CT images. All patients were treated by radiotherapy with curative intent; in 15 patients, adjuvant concomitant chemotherapy was administered. Mean follow-up time was 2.2 years. Actuarial (life-table) statistical analysis was done; multivariate analysis was performed using the Cox proportional hazards model.

Results

When the patients were stratified according to the median perfusion value (83.5 mL/min/100 g), those with the lower perfusion rate had a significantly higher local failure rate (p < 0.05). In the multivariate analysis, perfusion rate (p = 0.01) and T category (p = 0.03) were found to be the independent predictors of local failure. Perfusion rate had predictive value regarding neither regional control nor cause-specific survival.

Conclusion

CT-determined tumor perfusion rate was found to be an independent predictor of local outcome in irradiated head-and-neck cancer. The results of this study confirm the hypothesis that less-perfused tumors respond poorly to radiotherapy.

Introduction

Beginning with the observations of Gray et al. (1), oxygen is known to be a powerful radiosensitizer: Oxygen enhances the formation of free radicals or draws existing free radicals into chain reactions, producing new damaging free radicals; another mechanism postulated is that many irradiation-induced chemical changes are being blocked by the presence of oxygen.

An imbalance between oxygen supply and consumption, largely resulting from the presence of inadequate and heterogeneous vascular networks, leads to tumor hypoxia (2). In the second half of the twentieth century, studies have shown the clinically relevant effect of tumor hypoxia on the response to radiation therapy (2). Also in head-and-neck cancer, several studies found a significantly worse response to irradiation in tumors with a hypoxic subvolume 3, 4, 5. Recent laboratory and clinical data have shown that hypoxia is associated also with a more malignant phenotype, affecting genomic stability, apoptosis, angiogenesis, and metastasis (6). Direct quantification of tumor oxygenation can be expected to be of important prognostic value. Tumor oxygenation has been measured invasively using oxygen-sensitive needle electrodes in animal tumors and in certain human tumors. This technique can be used in head-and-neck tumors (5), but many primary tumors in this region are deeply seated, difficult to reach, and close to critical anatomic structures.

There is a need for a noninvasive method to measure tumor oxygenation, not only to predict outcome, but also to select patients for concomitant radiosensitizing therapy to overcome the hypoxia effect. Early efforts to overcome the limitations imposed by tumor hypoxia through use of carbogen breathing (7) or the radiosensitizer misonidazole (8) during radiotherapy in head-and-neck cancer have been unsuccessful. However, more recent studies have demonstrated that the oxygenation of head-and-neck tumors improves during carbogen breathing (9) or hyperbaric oxygenation (10). Increased oxygenation of tumors treated with carbogen and nicotinamide has been demonstrated in patients. Promising results have been obtained in nonrandomized clinical studies using this combination in conjunction with accelerated irradiation 11, 12.

Perfusion can be defined as the blood flow through a tissue of interest per unit of volume. Tumor perfusion and tumoral oxygen concentration are factors that are usually strongly linked, although tumor oxygenation depends also on oxygen consumption by the tumor cells. The oxygen availability or oxygen supply is the amount of oxygen carried by the blood to a given tissue per unit of time; it is the product of the perfusion rate and the arterial oxygen concentration.

The cross-sectional images obtained with CT provide detailed morphologic information. Besides an anatomic analysis, a functional analysis of the images is also possible. After i.v. bolus injection of an iodinated contrast agent, tissue and vessel attenuation changes can be observed during the first pass of this agent by rapid (“dynamic”) image acquisition at a given anatomic level. Time–density curves can then be constructed for observer-defined regions of interest (ROIs). Within the limits of some assumptions, tissue perfusion can be estimated based on the observed density changes: The time course of the iodine concentration is a measure of the regional perfusion, and this concentration is linearly correlated to tissue density, as seen on CT.

Several algorithms can be used to measure tissue perfusion with CT. The feasibility of the “gradient method” to measure perfusion in head-and-neck tumors was previously reported (13). This method can be performed during a routine CT study with little supplementary burden for the patient. Calculation of this parameter is relatively easy and inexpensive. A subsequent study showed an excellent intra- and interobserver reproducibility of the CT-determined perfusion values, a substantial intertumoral coefficient of variation (being larger than the intratumoral coefficient of variation), and absence of correlation between perfusion value and primary tumor volume. Furthermore, a tendency toward lower local control in cancers with a perfusion value smaller than the median was found (14).

The aim of the present study is to test the hypothesis that the outcome of patients with a low perfusion is worse than the outcome of patients with well-perfused tumors.

Section snippets

Methods and materials

Between March 1995 and January 2000, 114 patients with a primary squamous cell carcinoma in the head-and-neck region had a CT study of the head and neck that included a dynamic CT acquisition through the region of the tumor during bolus injection of a contrast medium (“perfusion study”). All these patients were treated with curative intent by radiotherapy. Two patients were excluded because treatment was interrupted as a result of intolerance; 7 patients were excluded because of problems during

Perfusion values

The mean CT-determined perfusion value was 88.8 mL/min/100 g (median 83.5, SD 46.5, range: 12.4–274.4 mL/min/100 g).

No correlation was found between primary tumor volume and perfusion rate (perfusion rate = −0.06 primary tumor volume + 27.9, R2 = 0.01).

Local control

The actuarial 2-year local control rate in the study population was 42%.

When the patients were stratified into two groups according to the median perfusion value, the patients in the high-perfusion group showed a significantly higher local

Discussion

This study demonstrates the value of CT-determined tumor perfusion as independent predictor of local control in head-and-neck cancer, treated by definitive radiotherapy, with or without adjuvant chemotherapy. Patients with a low perfusion value showed a statistically significantly higher local failure rate than those with a high perfusion value. Presumably, this is linked to more extensive and/or higher degree of hypoxia in low-perfused tumors, as suggested by others (18).

The perfusion rate was

References (31)

Cited by (175)

View all citing articles on Scopus
View full text