Abstract
Aim: To compare gross tumor volume (GTV) definition in locally advanced head and neck squamous cell carcinoma (LAHNSCC) using diffusion-weighted magnetic resonance imaging (DW-MRI) and computed tomography (CT) with intravenous contrast. Patients and Methods: Patients with LAHNSCC were imaged with CT and DW-MRI before treatment. GTV was delineated in both CT and DW-MRI images by two investigators. CT and MRI images were co-registered and volume data were extracted for statistical analysis. Results: In general, DW-MRI volumes [based on the apparent diffusion coefficient (ADC)] were smaller than CT-based volumes. For all patients, GTV delineation based on pre-treatment DW-MRI was significantly smaller than that based on CT scan (CT-GTV) (p=0.0078). The mean difference (95% limits of agreement) between the two investigators was −0.37 cm3 for CT-GTV and 0.17 cm3 for ADC-GTV measurements, respectively. Conclusion: DW-MRI radiotherapy GTVs are smaller than CT-based targets with less interobserver variability. Further validation of these preliminary results is necessary in a much larger patient group.
This is the era of concurrent chemoradiotherapy (CRT) for locally advanced head and neck squamous cell carcinoma (LAHNSCC), due to survival benefit of concomitant cisplatin-based chemotherapy and standard-fraction radiation over radiation therapy-alone (1). In parallel, this is the era of functional imaging techniques to assess tumor extension and estimate tumor response to a variety of anticancer treatments (2, 3). Magnetic resonance imaging (MRI), as well as fluorodeoxyglucose-positron emission tomography (FDG-PET), have been utilized extensively to better determine the location of tumor and are part of the routine work-up of patients with LAHNSCC (4).
Based on differences in the microscopic Brownian motion of water protons between tissues, quantified by a biological parameter – the apparent diffusion coefficient (ADC) – diffusion-weighted (DW) MRI is increasingly being used to assess changes in tissue characterization. Depending on tissues cellularity, DW-MRI provides information on the tissue microenvironment and tumor typically returns lower ADC than normal tissue due to its high cellularity (5).
The clinical use of intensity-modulated radiation therapy (IMRT) requires a precise delineation of target volumes and organs at risk (OARs), in order to ensure the best tumor coverage, sparing normal tissue (6).
Accuracy of target definition is paramount in radiation treatment planning, especially in the head and neck region because of its anatomical complexity. Traditionally, the gross tumor volume (GTV) is defined as evident disease and, nowadays, efforts to improve GTV delineation have been helped considerably with the evolution of new imaging modalities (7). The optimal choice of the imaging modality to accurately define the GTV in patients with LAHNSCC is still under debate. In IMRT, fusion of different imaging examinations during target volume definition is common but the use of DW-MRI has not been fully explored in this field (8).
DW-MRI has the potential to delineate a precise GTV. ADC values reflect the heterogenic organization of tumor tissue and ideally localize radioresistant areas. Thus, DW-MRI should be used at the contouring stage, in order to better delineate target volume, and consequently ensure highly individualized conformal treatment.
The aim of this study was to evaluate whether the use of DW-MRI could lead to delineation of a more accurate GTV compared to computed tomography (CT) scan with intravenous contrast in LAHNSCC. We correlated the GTV delineated on CT and DW images by two different investigators, and assessed if the agreement between investigators was improved by DW imaging.
Materials and Methods
Patient selection. This was a single-center observational study. It was approved by the Department of Radiological Sciences, Oncology and Pathology, Policlinico Umberto I Sapienza University of Rome and informed consent was obtained from all patients. The study protocol was registered in the registry of research projects of Sapienza University of Rome (no. C26N14N9H3).
Patients with histologically proven HNSCC were included, provided they were ≥18 years of age, had performance status ≤2, had adequate renal, hepatic, and bone marrow function, had T3-4 tumor with/without positive lymph nodes, without any evidence of distant metastases by imaging modalities.
The major exclusive criteria was contraindication to MRI examination. Patients were excluded from the study in the case of synchronous tumors, severe comorbidities (cardiovascular disease, history of neurological or psychiatric disorders, collagenosis) or previous radiotherapy to the head and neck region.
Clinical examinations including nasopharyngolaryngoscopy were combined with radiologic imaging, CT and MRI, to assess the precise local (T), regional nodal (N), and distant (M) extent of the tumor. Patient tumors were classified according to the American Joint Committee on Cancer tumor, nodes, metastasis (TNM) Staging System (9). Preventive dental care, as well as nutritional evaluation, occurred 15 days before CRT.
Treatment plan. All patients were treated with definitive CRT. Radiation therapy was delivered with IMRT technique at a dose of 67.5 Gy (2.25 Gy/fraction) to macroscopic disease, 60 Gy (2 Gy/fraction) to high risk regions and 54 Gy (1.8 Gy/fraction) to sites of potential disease, with 6-15 MV energy photons. Chemotherapy consisted of two courses of cisplatin infusion at 100 mg/m2, on day 1 and 22 of the course of IMRT.
Planning CT imaging. Head and neck CT scan with intravenous contrast was acquired for each patient, one week prior to CRT. Patients were treated supine and were immobilized in a thermoplastic shell, fixed to the couch in five places. Images were acquired with 3 mm slice thickness and were obtained from base of skull to the top of the aortic arch.
Magnetic resonance imaging. All patients were examined by MRI shortly after simulation CT scans. All MRI examinations were performed with a 3.0 T MR system (Discovery 750; GE Healthcare, Milwaukee, WI, USA) in the treatment position. All patients underwent DW imaging in addition to the conventional head and neck MRI protocol (axial T2-weighted, axial and coronal T1-weighted, contrast-enhanced axial and coronal fat-suppresses T1-weighted, and coronal short tau inversion recovery sequences) (10). Axial DW images were obtained using the single-shot echo-planar imaging technique. Diffusion-encoding gradients were applied as a bipolar pair at 2 b-values of 0 and 1,000 s/mm2. ADC maps were generated from DW images from the following equation on the systems workstation: ADC=−ln (Sb/S0)/b, where Sb was the mean signal intensity with diffusion weighting b, S0 was the mean signal intensity for a b-value of 0 s/mm2.
Image co-registration. All images were registered to the planning CT performed in the treatment position before CRT. Co-registrations were performed using a deformable fusion method. DW-MRI images were downloaded from the institutional database. Each co-registration was performed within 3 mm between CT and DW-MRI. The procedure has been validated for the head and neck region and found to be accurate and reproducible.
GTV delineation. For radiation treatment plan, the GTV was delineated based on CT and DW-MRI. The GTV contours were manually delineated by two radiation oncologists (FDF, DM) specialized in the head and neck region.
For CT scan, GTV contours were obtained from contrast phase images (CT-GTV) using standard CT window levels without reference to MRI data sets. For DW-MRI, GTV delineations were performed on the ADC map generated from DW imaging (ADC-GTV), but information provided by the pre- and post contrast T1-weighted images and the T2-weighted images were used when felt relevant.
For each CT-GTV and ADC-GTV, the volume in cubic centimeters was recorded.
Statistical analysis. Statistical analysis was performed using the RStudio-0.98.1091 software. Standard descriptive statistics were used to evaluate the distribution of each potential factor. Continuous data are given as the median (range), and categorical data as the number of observations and ratios.
The GTVs obtained from CT and DW-MRI were compared and the Wilcoxon signed-rank test was used to determine whether significant differences existed between the matched samples. Bland-Altman analysis was used to assess interobserver agreement (mean difference, 95% limits of agreement). All reported p-values are two-sided, and p-values lower than 0.05 were considered significant.
Results
Patients' characteristics. Eight consecutive patients with LAHNSCC treated with primary CRT were included in the study between November 2014 and May 2015. Baseline patient and primary tumor characteristics are reported in Table I. The primary sites were the oropharynx (n=5; 62.5%), hypopharynx (n=1; 12.5%) and larynx (n=2; 25%). Patients were classified as having stage III-IVa tumor with no evidence of metastatic disease (M0).
All patients received concomitant platinum-based chemotherapy and were treated using IMRT technique.
GTV characteristics. Data on CT-GTV and ADC-GTV measurements are shown in Table II. In total 16 image sets (one CT and one DW-MRI for each of the eight patients) were analyzed by each investigator. No significant artifacts were recorded. ADC-GTVs were smaller and statistically different from CT-GTVs (p=0.0078 investigator 1; p=0.0078 investigator 2). Overall, the mean ADC-GTV was less than the CT-GTV. The mean reduction of CT-GTV to ADC-GTV was 6.5±5.7 cm3. Using ADC-GTV for IMRT treatment planning resulted in a reduction of the high-dose GTV in 100% of cases. A graphic comparison of the differences in GTV sizes obtained from CT and DW-MRI (ADC) for each patient included in the study is displayed in Figure 1.
Patient and tumor characteristics.
Mean gross tumor volume (GTV) obtained by the two investigators for the images by computed tomography (CT) and diffusion-weighted magnetic resonance imaging with apparent diffusion coefficient (ADC).
Comparison of gross tumor volume (GTV) sizes obtained from computed tomographic (CT) and diffusion-weighted magnetic resonance imaging using apparent diffusion coefficient (ADC) for each patient, by investigator 1 (#1) and investigator 2 (#2).
The mean difference (95% limits of agreement) between the two investigators was −0.37 (−1.68 to 0.93) cm3 for CT-GTV measurements (Figure 2) and 0.17 (−0.26 to 0.61) cm3 for ADC-GTV measurements (Figure 3). The intraclass correlation coefficient was 0.44 for ADC volumes while that for CT-obtained volumes was 1.31.
Discussion
In this study, we found that primary tumor volume was more directly identifiable on DW-MRI compared to CT. We demonstrated that the use of DW-MRI co-registration for the delineation of primary LAHNSCC led to significantly smaller GTVs than those delineated on CT (p=0.0078). Moreover, DW-MRI reduced interobserver variation, with 95% limits of agreement slightly stricter for the ADC-GTV versus CT volumes, and it was possibly a reflection of the greater ability to interpret ADC images.
These results suggest that the use of ADC-GTV would be a safe approach in the era of IMRT planning, reducing uncertainty in delineating tumor volume and consequently geographical miss. With its higher functional ADC contrast, DW-MRI enabled the design of precision GTV, optimizing dose delivery to target volume and respecting dose constraints to OARs. This would facilitate relative dose sparing to the normal tissues, especially if a extra-dose is given based on the GTV.
Head and neck region OARs include parotid and submandibulary glands, constrictor muscles and mandible, principally. These organs have dose-dependent toxicities (11). Osteoradionecrosis may be considered the predominant dose-limiting toxicity for LAHNSCC, whereas xerostomia and dysphagia are dose- and volume-dependent, and remain a major challenge for radiation oncologists and patients (12-14). Co-registration of MRI using ADC analysis in radiotherapy treatment planning could contribute to reduce irradiation of OARs and facilitate dose escalation strategy.
Bland Altman plot (95% limits of agreement) of gross tumor volume obtained from computed tomography for the two investigators. The x axis is the mean of two measurements each; the y axis is the difference between them.
Diagnostic MRI has been shown to be of great importance in loco-regional staging of LAHNSCC because of better spatial resolution and soft-tissue contrast (15). Functional imaging techniques, such as DW-MRI, depict biological activity and tumor structural characteristics, scoring a sensitivity and specificity of approximately 95% (16). While DW-MRI has been considered a potential biomarker for predicting treatment response, even if without definitive results, its employment in the target volume definition is still inconsistent due to the intrinsic limitations of the head and neck region in performing MRI acquisitions (17-19). However, DW-MRI can provide better definition of primary tumor extent and can easily reduce CT image distortion and artifact from metal prosthesis. Thus combining the two modalities could improve the therapeutic index of radiotherapy for LAHNSCC and our analysis showed that DW-MRI provided smaller and more accurate GTVs than does CT, reflecting its potential role in the treatment planning stage.
In the literature, several studies have described the feasibility of using DW-MRI to define treatment targets in other primary tumor sites, including gastrointestinal tract and gynecological region (20-22). Dalah et al. studied the variability of GTV and OARs delineation in RT planning for pancreatic cancer. GTVs were defined using different multimodality imaging, such as CT, PET and DW-MRI. A total of 19 patients were analyzed and GTVs defined from DW-MRI demonstrated better agreement with pathological specimen than other imaging methods (20). Regini et al. compared the rectal GTV on T2-weighted and DW-MRI images in 27 consecutive patients with rectal cancer. GTVs were delineated by two different observers. DW-MRI permitted smaller GTV delineation, but overall observed agreement was not improved (21). Esthappan et al. examined the feasibility of using DW-MRI for target delineation in brachytherapy of cervical cancer compared against T2-weighted imaging. Based on 15 patients' data, ADC-GTV gave generally smaller than T2-weighted delineated volumes. Therefore, independently of tumor location, the authors suggested incorporating DW-MRI into routine GTV delineation (22).
Bland Altman plot (95% limits of agreement) of gross tumor volume by diffusion-weighted magnetic resonance imaging using apparent diffusion coefficient for the two investigators. The x axis is the mean of two measurements each; the y axis is the difference between them.
This study is clearly limited by the small patient numbers. However, despite the small sample size, statistically significant differences existed between GTV measurement from different imaging modalities. Contouring applications of DW-MRI should gain increasing importance, including in the head and neck region. Further investigations with a larger population are warranted to strengthen our results. However, it is important to realize that DW-MRI is not expensive, takes only a few minutes and does not need intravenous contrast.
In addition, another potential limitation of our study is the lack of a comparative analysis with surgical specimens but this is related to the curative intent of the primary treatment.
Briefly, in daily practice, DW-MRI could be added to routine imaging studies of LAHNSCC to improve accurate definition of the target volume. Based on PET studies attempting to identify hypoxic sub-volume within the GTV to allow the use of higher doses in hypoxic cells (7), the change in ADC could have an accurate role in identifying foci of a more aggressive tumor phenotype and it could be useful for modulating radiation. By use of the ADC evolution to dose-paint GTV, a higher dose could be administered to areas of ADC decrease, while a lower dose could be given to normal ADC tissue. As real-time treatment planning becomes possible, the ADC value could be used to conform radiation to the shape of the ADC measurements in order to provide a more adequate target coverage. It can be hoped that as the ADC decreases, increased effectiveness resulting from higher doses will be obtained. For future research, the next step is to investigate whether DW-MRI of LAHNSCC could be used to modulate the dose distribution predicting infield tumor control during definitive CRT.
Our study, although exploratory in nature, strongly suggests that DW-MRI may improve tumor delineation and reduce tumor GTV compared to CT. Whether the use of ADC-GTV would translate into improved patient outcomes remains a hypothesis that would have to be tested. A long-term analysis, as well as a larger sample size, is needed to draw definitive conclusions.
DW-MRI seems an optimal candidate for radiotherapy planning. The incorporation of fused DW-MRI images may reduce uncertainty in GTV delineation and reduce inter-investigator variation.
Conceptually, the role of multimodality imaging in radiotherapy planning to approach LAHNSCC target definition should be paramount.
- Received May 20, 2016.
- Revision received June 11, 2016.
- Accepted June 13, 2016.
- Copyright© 2016 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
