Pretreatment and early intratreatment prediction of clinicopathologic response of head and neck cancer to chemoradiotherapy using 1H-MRS

Authors

  • Ann D. King FRCR,

    Corresponding author
    1. Department of Diagnostic Radiology & Organ Imaging, Faculty of Medicine, Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, New Territories, Hong Kong S.A.R. China
    • Department of Diagnostic Radiology and Organ Imaging, Faculty of Medicine, Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, 30-32 Ngan Shing Street, New Territories, Hong Kong S.A.R. China
    Search for more papers by this author
  • David K.W. Yeung PhD,

    1. Department of Diagnostic Radiology & Organ Imaging, Faculty of Medicine, Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, New Territories, Hong Kong S.A.R. China
    Search for more papers by this author
  • Kwok-hung Yu FRCR,

    1. Department of Clinical Oncology, Faculty of Medicine, Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, New Territories, Hong Kong S.A.R. China
    Search for more papers by this author
  • Frankie K.F. Mo PhD, MD,

    1. Department of Clinical Oncology, Faculty of Medicine, Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, New Territories, Hong Kong S.A.R. China
    Search for more papers by this author
  • Kunwar S. Bhatia FRCR,

    1. Department of Diagnostic Radiology & Organ Imaging, Faculty of Medicine, Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, New Territories, Hong Kong S.A.R. China
    Search for more papers by this author
  • Gary M.K. Tse FRCPC,

    1. Department of Anatomical and Cellular Pathology, Faculty of Medicine, Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, New Territories, Hong Kong S.A.R. China
    Search for more papers by this author
  • Alexander C. Vlantis FRCS,

    1. Department of Otorhinolaryngology, Head and Neck Surgery, Faculty of Medicine, Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, New Territories, Hong Kong S.A.R. China
    Search for more papers by this author
  • Jeffrey K.T. Wong FRCR,

    1. Department of Diagnostic Radiology & Organ Imaging, Faculty of Medicine, Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, New Territories, Hong Kong S.A.R. China
    Search for more papers by this author
  • Chen-wen Hu FRCR,

    1. Department radiology Shenzhen Fifth Hospital and Luo-hu District Hospital, Guang-dong Province, People's Republic of China
    Search for more papers by this author
  • Anil T. Ahuja FRCR

    1. Department of Diagnostic Radiology & Organ Imaging, Faculty of Medicine, Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, New Territories, Hong Kong S.A.R. China
    Search for more papers by this author

Abstract

Purpose:

To determine if choline (cho) identified by proton magnetic resonance spectroscopy (1H-MRS) performed pretreatment and early in the course of treatment predicts clinicopathologic response of head and neck squamous cell carcinoma (HNSCC).

Materials and Methods:

In all, 60 patients with HNSCC scheduled to undergo concurrent chemoradiotherapy or radiotherapy alone were recruited. 1H-MRS was performed pretreatment and early intratreatment (2 weeks after start of treatment). Cho:creatine and cho:water ratios at each timepoint and change in the ratios between the two timepoints were correlated with locoregional failure, distant metastases, overall survival, and cancer-related death. Statistical analysis was performed using logistic regression and chi-square and a P-value of < 0.05 was considered statistically significant.

Results:

Cho was identified in 47/49 successful pretreatment spectra and 42 of these 47 underwent successful 1H-MRS early intratreatment, of which 21 showed persistent cho. Locoregional failure occurred in 15, distant metastases in 6, and death in 15 patients; the follow-up period in survivors ranged from 13–64 months (mean, 39 months). No statistically significant correlation was found between 1H-MRS parameters and clinical endpoints.

Conclusion:

The pretreatment cho and change in cho early during a course of treatment did not predict clinical outcome. J. Magn. Reson. Imaging 2010;32:199–203. © 2010 Wiley-Liss, Inc.

CHEMORADIOTHERAPY (CRT) and radiotherapy (RT) are nonsurgical treatments for patients with head and neck squamous cell carcinoma (HNSCC) and for advanced stage disease CRT has been gaining popularity as a viable alternative to surgery. However, not all patients respond to CRT/RT; thus, early identification of nonresponders would allow CRT/RT regimens to be modified or changed to surgery. Several markers are used in clinical practice for the pretreatment prediction of clinical outcome, including tumor staging, tumor volumes, histopathology and molecular biomarkers, but all have limitations, and at present there are no intratreatment markers for predicting eventual treatment response.

Choline (cho) measured within a tumor pretreatment and change in cho levels early in the course of treatment have the potential to act as new predictive markers. Cho containing compounds are identified in vivo as a spectral peak at 3.2 ppm using proton magnetic resonance spectroscopy (IH-MRS), which is a technique that can be added easily to the routine magnetic resonance imaging (MRI) examination. Cho has been identified in a range of head and neck cancers including HNSCC (1–8), where pretreatment cho levels correlate with histological grading (7) and posttreatment cho levels with the presence of residual/recurrent cancer after surgery or chemoradiotherapy (8). Changes in cho ratios from pretreatment to posttreatment also have been reported. In one small series, all five patients with successfully treated HNSCC showed a reduction in cho:water ratio of the tumor (5), while in our previous study persistence of cho 6 weeks after treatment was a marker for residual cancer (9). With regard to the prediction of patient outcome, there is a promising study by Bezabeh et al (4) that examined primary HNSCC biopsy specimens taken before treatment, which was mainly surgical-based. That study found significantly higher cho:creatine (cr) ratios in the tumors of patients with a poor response in terms of tumor recurrence and survival.

At present, to the authors' knowledge, there have been no in vivo IH-MRS studies evaluating the potential of using pretreatment cho ratios and early intratreatment change in cho ratios for predicting outcome in HNSCC. The aim of the study, therefore, was to determine whether the pretreatment cho ratios and early intratreatment induced changes in cho ratios could be used as a marker to predict treatment response using the clinical endpoints of locoregional failure in the neck, distant metastases, overall survival, and cancer-related death.

MATERIALS AND METHODS

Patients

Patients planned for primary treatment with concurrent CRT or RT with curative intent were enrolled in this prospective study if they had biopsy-proven, previously untreated, Stage III or IV squamous cell carcinoma in the head and neck according to the 1997 AJCC staging system. Patients were required to provide written informed consent before study entry. The study protocol was approved by the Institutional Review Board. 1H MRS was performed on the primary tumor or on a cervical lymph node metastasis larger than 1 cm3. The primary tumor was chosen preferentially for MRS examination, but for those patients where it was judged likely that the MRS would fail at the primary site because of technical difficulties (small size, irregular shape, artifact caused by susceptibility effects, and movement), the largest solid node was chosen instead. Histology of the biopsy specimen was the reference standard for the primary tumor, while the reference standard for a metastatic node was either histology or imaging criteria (shortest axial diameter ≥11 mm in the jugulodigastric region and ≥10 mm in all other regions of the neck).

MRI and 1H MRS

MRI and 1H MRS were performed on a 1.5 T whole-body system (Intera-NT, Best, Netherlands) with a 30 mT/m maximum gradient capability. A standard volume head and neck coil was used to localize the lesion for 1H MRS and conventional MRI was performed for clinical staging and follow-up examinations. For optimal signal reception, 1H MRS was performed using a 20-cm diameter circular receive-only surface coil placed over the lesion of interest. From each volume of interest (VOI), a water-suppressed spectrum was acquired using the point-resolved spectroscopic (PRESS) sequence (TR/TE 2000/136 ms). Water suppression was achieved using a selective inversion recovery technique and 64 water-suppressed signals were acquired at a spectral bandwidth of 1000 Hz. An unsuppressed water signal with 16 averages was acquired as reference spectrum. Data were exported and processed on an offline computer (Dell Precision 650 Workstation).

Amplitude and line width of spectral peaks were determined using a time-domain fitting program (AMARES, implemented in the MRUI software package) (10). As the signal-to-noise ratio of head and neck spectra may be relatively low, for a peak to be considered present its amplitude should be at least twice the standard deviation of the baseline noise. Manually selected resonance frequency and line width of cho and cr peaks were used as the starting values. Details of prior knowledge and standard procedures used to remove residual lipids and water have been described elsewhere (9). Analysis of the spectra was made by a physicist with more than 11 years of experience in spectroscopy who was blinded to the clinical results.

Calculated cho, cr, and water peak amplitudes were used to determine the cho/Cr and cho/water ratio for each lesion. MRI and 1H MRS were performed pretreatment and 2 weeks after the start of a course of treatment. Pretreatment and early intratreatment for the following 1H MRS parameters were evaluated: cho:cr ratio, cho:water ratio, absolute and percentage change between pretreatment and early intratreatment ratios, and persistence of cho on the early intratreatment scan.

Clinical Endpoints

There were four clinical endpoints; locoregional failure in the HNSCC site examined by IH-MRS, distant metastases, overall survival, and cancer-related death. Locoregional failure was determined by histological confirmation of HNSCC (biopsy or surgical resection) or by a serial increase in size at any time after the end of treatment. Locoregional control was determined by histological confirmation (surgical resection) or by follow-up where a residual lesion remained static or decreased in size (period of follow-up at least 1 year after the end of treatment).

Statistical Analysis

Statistical analysis was performed using SAS software v. 8.2 (Cary, NC). The cho:water and cho:cr ratios pretreatment and early intratreatment and the change in ratios between these two timepoints were correlated with locoregional failure, distant metastases, and death using logistic regression. The persistence of cho on the early intratreatment examination was correlated with the same clinical endpoints using the chi-square test. A P-value of < 0.05 was considered statistically significant.

RESULTS

Patients and 1H MRS

In 60 patients who underwent concurrent CRT or RT the pretreatment 1H MRS failed in 11 patients and in two patients the examination was successful but no cho was detectable in the tumor. The study was therefore comprised of 47 patients (43 males and four females, mean age 57 years, range 43–74 years) who presented with a primary carcinoma of the hypopharynx/larynx (20), oral cavity/oropharynx (22), paranasal sinus/nasal cavity (3), and esophagus (2). Patients underwent concurrent CRT (n = 43) or RT alone (n = 4). 1H MRS was performed at 22 primary sites (47% of patients) and 25 neck nodal sites (53% of patients). The size of the nodes studied by MRS were all well above the criterion for a metastatic node (minimum diameter range 1.5–6.0 cm, median 2.7 cm; maximum diameter range 1.8–10 cm, median 3.7 cm). Pretreatment cr peaks could not be identified in 15 patients. Pretreatment cho:water and cho:cr ratios were obtained in 47 and 32 patients, respectively. The early intratreatment 1H MRS failed in five patients, leaving 42 patients with successful spectra. In these 42 patients a cho peak was detected at 2 weeks in 21 patients and had resolved in 21 patients. Of the 21 patients with detectable cho, the cr peak could only be detected in 13. The mean, standard deviation, and range of the cho:water and cho:cr pretreatment and early intratreatment of the whole cohort of patients is shown in Table 1. Change in cho:water ratios from pretreatment to early treatment showed a rise (n = 8), fall (n =13), or resolution (n = 21) (Fig. 1a,b). Change in cho:cr ratios from pretreatment to early treatment showed a rise (n = 2), fall (n =11), or resolution (n = 21).

Figure 1.

1H MRS spectrum from a metastatic node in a patient with locoregional control after treatment. Pretreatment the node (arrows) shown on the T1-weighted image (a) shows a cho peak with a cho:water ratio of 2.3 × 10−3 on MRS (b). Early intratreatment the node (arrows) shown on the T1-weighted image (c) shows resolution of the cho peak on MRS (d).

Table 1. Choline Ratios Pretreatment and Early Intratreatment
 Pretreatment cho:waterPretreatment cho:crEarly intratreatment cho:waterEarly intratreatment cho:cr
No. patients47322113
Mean1.29 × 10−32.311.08 × 10−32.00
Range0.15 −9.52 × 10−31.08 – 4.370.17– 5.96 × 10−30.83- 3.53
Standard deviation1.61 × 10−30.831.47 × 10−30.81

Clinical Endpoints

Five of the 47 patients died within the first year before the nature of a residual neck mass was determined. In the remaining 42 patients locoregional failure occurred in 15 patients (36%) (range 1–16 months, mean 6 months after treatment). Twenty-seven patients had locoregional control based on surgical excision in seven and clinical follow-up in 20 (range 13–64 months, mean 36 months after treatment). The relationship between locoregional failure and changes in the cho:water and cho: cr ratios between the pretreatment and early treatment DWI are shown in Table 2. One patient had distant metastases at diagnosis and five patients developed distant metastases after treatment (range 1–10 months, mean 5 months). Noncancer-related deaths occurred in three patients and in the remaining 44 patients, 15 (34%) died (range 3–26 months, mean 12 months) and 29 (66%) survived (range 13–64 months, mean 39 months).

Table 2. Change in Choline Ratios Between Pretreatment and Early Intratreatment in Relationship to Locoregional Failure
  Total number patientsNumber of patients with locoregional failureNumber of patients with locoregional control
Cho:waterRise835
 Fall/resolution301119
Cho:crRise220
 Fall/resolution17611
ChoPersists19118
 Resolved18612

Statistical Analysis

There were no statistically significant correlations between absolute pretreatment or early intratreatment cho:cr or cho:water ratios, or change in ratios (absolute or percent) between these timepoints, with the clinical endpoints of locoregional failure, distant metastases, overall survival, or cancer-related death (P values ranged between 0.2 and 0.84 using logistic regression). In addition, the persistence of cho by 2 weeks intratreatment was not significantly related to the four clinical endpoints (P values ranged between 0.4 and 1.0 using chi-square test).

DISCUSSION

In vivo 1H MRS identifies elevated cho peaks in a wide range of cancers in the brain and body, and as such cho has become an important focus for research in oncology. The cho peak identified by 1H MRS is related to the synthesis and breakdown of cell membranes and is formed by a combination of free choline, phosphocholine, and glycerophospocholine. Cells that acquire a malignant phenotype show an increase in the cho-containing compounds (12), and cho values reflect cellular density (13) and proliferative activity (14). These attributes make cho an attractive potential marker for predicting and evaluating cancer treatment response. Early results from the in vivo study of cancers outside the head and neck have been promising. Elevated pretreatment cho has been shown to correlate with a higher tumor grades (15, 16), and falls in cho, during or after a course of treatment, have been demonstrated in a range of cancers. In general, tumors that show a greater fall or disappearance of cho have been shown to have a better response to treatment, while static or even rising cho level are associated with a poorer response (17–20). The hypothesis in this study was that clinical responders would show lower pretreatment cho levels and greater falls in cho early in the course of treatment. This hypothesis was tested in this study using clinicopathological endpoints with a mean of 3 years follow-up in survivors.

Unfortunately, no significant correlation was found between the pretreatment cho ratios of HNSCC and clinical outcome in terms of locoregional disease, distant metastases, overall survival, or death. Therefore, performing 1H MRS on the in situ tumor in a clinical environment could not reproduce the results of Bezabeh et al (4) obtained from tumor biopsies. The results may in part reflect the difficulties of performing 1H MRS in the head and neck, although a similarly nonsignificant result also has been shown in a breast cancer study correlating pretreatment cho with response to neoadjuvant chemotherapy (20).

Early intratreatment changes had even greater potential to predict eventual clinical response. Magnetic resonance markers using other functional MRI techniques, such as dynamic contrast enhancement in breast cancers (21) and diffusion-weighted imaging in head and neck cancers (22), have shown promise in the early prediction of treatment response. In this HNSCC study tumors with a greater fall or a disappearance in cho on 1H MRS were expected to show a better outcome. However, once again, no significant correlation could be found. Two weeks after the start of treatment the cho peak had already disappeared in half of the tumors and the remaining tumors showed both rising and falling cho ratios, which did not ultimately correlate with any of the clinical endpoints.

The failure rate of the MRS was relatively high in this clinical setting. Patient motion and susceptibility artifact caused most of the failures in patients who were initially recruited for the study but then excluded from analysis because of an unsuccessful baseline study. Two weeks after the start of CRT there were further failures because treatment side effects meant that some patients were too ill to attend or tolerate the scan. In addition, lower spectral resolution from the shimming difficulties, which is inherent in the head and neck region, also degraded the spectral peaks, causing difficulty especially with detection of cr. As a result of these problems, although 60 patients were recruited the eventual numbers for analysis was smaller, which was one of the limitations of the study. In this study both the primary and nodal sites were combined in the analysis and a much larger study would be needed to determine if there is any difference in the results between these two sites and if there are any differences in results from different primary sites along the aerodigestive tract. The technique also has other in-built limitations: the size of the VOI for 1H MRS could not be smaller than 1 cm3, making it insensitive to small lesions, while on the other hand not all regions in a large or irregular shaped lesion could be investigated. Finally, a further limitation may lie with the timing if the early intratreatment scan. The 2-week assessment point may have been too late in the treatment cycle to perform a predictive 1H MRS. A breast cancer study has shown neoadjuvant chemotherapy-induced cho changes within 24 hours of the start of treatment predict response (17). Advancing the intratreatment scan to the first few days after the start of treatment may improve the predictive ability of 1H MRS, but further research will be required to determine the optimum time for early assessment of intratreatment response in HNSCC.

In conclusion, the head and neck remains a challenging region in which to perform 1H MRS and inevitably some of the examinations will fail. Previous studies have shown the potential of 1H MRS for the assessment of residual/recurrent cancer in HNSCC, but this current study was unable to establish any predictive value of 1H MRS. The cho:water and cho:cr ratios pretreatment and the intratreatment change in these ratios 2 weeks after the start of treatment were not predictors of clinicopathological response. Nevertheless, 1H MRS remains a promising area and it is hoped that improvements in 1H MRS techniques implemented on higher field strength scanners and further research into the optimal timing of the intratreatment examination coupled with larger studies with more emphasis on cancer subsets eventually will prove more fruitful.

Ancillary