Original article Utility of FMISO PET in advanced head and neck cancer treated with chemoradiation incorporating a hypoxia-targeting chemotherapy agent Rodney J. Hicks1, 2, Danny Rischin2, 3, Richard Fisher4, David Binns1, Andrew M. Scott5, Lester J. Peters6 1

Centre for Molecular Imaging, Peter MacCallum Cancer Centre, Locked Bag No 1, A’Beckett St., Melbourne, 8006, Australia Department of Medicine, St Vincent’s Medical School, University of Melbourne, Melbourne, Australia 3 Division of Haematology and Medical Oncology, Peter MacCallum Cancer Centre, Melbourne, Australia 4 Centre for Biostatistics and Clinical Trials, Peter MacCallum Cancer Centre, Melbourne, Australia 5 Centre for PET, and Ludwig Institute for Cancer Research, Austin Hospital, Melbourne, Australia 6 Division of Radiation Oncology, Peter MacCallum Cancer Centre, Melbourne, Australia 2

Received: 10 April 2005 / Accepted: 5 June 2005 / Published online: 26 August 2005 © Springer-Verlag 2005

Abstract. Purpose: The purpose of the study was to evaluate [18F]fluoromisonidazole (FMISO) PET in advanced head and neck cancer during hypoxia-targeting therapy. Methods: Fifteen of 16 patients in a phase I trial of chemoradiation plus tirapazamine (specific cytotoxin for hypoxic cells) in advanced (T3/4 and/or N2/3) head and neck cancer underwent serial [18F]fluorodeoxyglucose (FDG) and FMISO PET. We have previously reported excellent early clinical outcome of these patients and now review FMISO PET results in the context of longer followup of this patient cohort. Results: Based on blinded qualitative scoring by two readers, FMISO PET was positive in 13/15 patients at baseline: 12/15 of primary sites and 8/13 neck nodes were scored as positive. All sites of corresponding FDG and FMISO abnormality at baseline showed marked qualitative reduction of uptake within 4 weeks of commencing therapy, consistent with effective hypoxia-targeted therapy. With a median follow-up of 6.9 years, there have been only four locoregional failures, while three other patients have died of metachronous lung cancer. The 5-year overall survival was 50% (95% CI 27–73%), the 5-year failurefree survival was 44% (95% CI 22–68%) and the 5-year freedom from locoregional failure was 68% (95% CI 38–88%). Conclusion: The high prevalence of hypoxia demonstrated on FMISO PET imaging is consistent with the advanced disease stage of these patients and would be expected to predict an adverse prognosis. Evidence of

.

Rodney J. Hicks (*) Centre for Molecular Imaging, Peter MacCallum Cancer Centre, Locked Bag No 1, A’Beckett St., Melbourne, 8006, Australia e-mail: [email protected] Tel.: +61-3-96561852, Fax: +61-3-96561826

the early resolution of FMISO abnormality during treatment, associated with excellent locoregional control in this patient cohort, supports further investigation of hypoxia-targeting agents in advanced head and neck cancer. Keywords: PET – Advanced head and neck cancer – Chemoradiotherapy – Hypoxia Eur J Nucl Med Mol Imaging (2005) 32:1384–1391 DOI 10.1007/s00259-005-1880-2

Introduction Although the addition of concurrent chemotherapy to radiation has improved the results in head and neck cancer [1], the prognosis for patients with locoregionally advanced disease remains poor. One of the potential factors in the poor outcome of advanced head and neck cancers is the presence of tumour hypoxia. Low tissue oxygenation has long been recognised to increase resistance to radiation [2] and its presence has been documented to be an adverse prognostic factor in squamous cell carcinoma of the head and neck region [3–6]. One suggested approach to improving response rates and survival in advanced head and neck cancer has been to target hypoxic cells by combining a hypoxic cytotoxin with radiation. Tirapazamine is a bioreductive benzotriazine compound that demonstrates differential toxicity for hypoxic cells [7]. We have previously reported clinical results of a phase I trial [8] which evaluated a novel combined chemoradiation protocol that sought to exploit both the hypoxic cytotoxicity of tirapazamine [9] and its potentiation of cisplatin cytotoxicity [10] in a patient group with a high likelihood of tumour hypoxia. As part of this trial we performed serial

European Journal of Nuclear Medicine and Molecular Imaging Vol. 32, No. 12, December 2005

1385

[18F]fluorodeoxyglucose (FDG) and hypoxia imaging with [18F]fluoromisonidazole positron emission tomography (PET) scans. Imaging of hypoxia has practical advantages compared with the more established technique of polarographic oxygen probe measurement. These advantages include lack of invasiveness, evaluation only of viable regions within tumours, potentially wider availability and the ability to detect hypoxia in deep-seated lesions that might be inaccessible to probe insertion. The ability to perform serial evaluation to determine whether tumour hypoxia resolves during treatment may also be of benefit. In this article we report long-term outcomes in a cohort of patients who, based on paired, blinded qualitative reading of FMISO PET, demonstrated a high prevalence of presumed hypoxia and its resolution with hypoxia-targeting therapy.

Materials and methods Eligibility To be eligible for treatment using this protocol, patients were required to have histologically proven squamous cell carcinoma of the head and neck with a bulky (T3/4) primary, advanced (N2/3) nodal metastasis or both. At least one bi-dimensionally measurable lesion was required for computed tomography (CT) monitoring and no systemic metastatic disease (M0) could be present on conventional staging. Systemic metastases detected on FDG PET that could not be verified by correlative imaging or biopsy did not make the patient ineligible since a “benefit of the doubt” philosophy was adopted. Other eligibility criteria have previously been described [8]. Written informed consent was obtained from all patients and our institutional ethics committee approved the protocol.

Treatment protocol The treatment schedule and dose-limiting toxicity data have also been described previously [8]. In summary, tirapazamine and cisplatin were given intravenously immediately prior to radiation on day 2 of weeks 1, 4 and 7. Additional doses of tirapazamine were administered three times weekly immediately prior to radiotherapy without cisplatin in weeks 2, 3, 5 and 6 in the first six patients and restricted to weeks 2 and 3 in subsequent patients. Radiation therapy consisted of 70 Gy in 35 fractions over 7 weeks. The radiation was given via a shrinking field technique encompassing the gross clinical disease and sites suspected of harbouring sub-clinical disease, including normal-sized nodes visualised on FDG PET scanning. Patients with a residual neck mass at 12 weeks, who had achieved a complete response at the primary site, underwent a neck dissection.

Pretreatment and follow-up imaging Before enrolment and 12 weeks following the end of radiotherapy, all patients underwent a helical CT (X-press SX, Toshiba Corp, Japan) of the head and neck with dynamic contrast. PET with FDG was performed at baseline, mid-treatment (week 4) and 12 weeks after completing treatment (week 19). All FDG PET studies were performed after a fast of at least 6 h. Immediately prior to FDG administration, patients were sedated with 10 mg IV diazepam to reduce physiological uptake of FDG in skeletal muscle and brown

fat. Imaging commenced 60 min after administration of approximately 120 MBq FDG using a dedicated PET scanner operating in 3-D acquisition mode (PENN-PET 300H, UGM Medical System Inc, Philadelphia PA). On each occasion PET images of the neck were acquired and processed using measured attenuation correction and iterative reconstruction [11] with the patient positioned as for radiotherapy, with arms by the side and neck extended. A contour colour-scale allowing variable upper and lower count thresholds to be outlined was used to define the body outline, brain uptake and tumoural uptake on the baseline FDG PET scan, and these landmarks were used to align and interpret the subsequent PET scans. The alignment was performed using co-registration software, provided by the scanner manufacturer, that allows visual translation and rotation of the image sets. Since all scans were acquired on the same scanner in the treatment position, no warping or scaling was required. Qualitative evaluation and determination of maximal standardised uptake values (SUVmax) were performed blinded to CT results. For subsequent scans the tumoural region of interest (ROI) defined on the baseline study was re-applied to the co-registered image set to ensure comparability of ROI assignment. On the baseline and 12week post-treatment scans, a screening study of the thorax and upper abdomen was also performed. This study was generally performed without attenuation correction. FDG PET lesions were scored qualitatively by consensus of two readers blinded to CT results, according the following scheme: 0=uptake less than background 1=no regions of focal uptake greater than background 2=focal uptake mildly greater than background 3=focal uptake moderately greater than background 4=focal uptake markedly greater than background A score of 2 or above was considered positive. Background was generally taken to be paravertebral skeletal muscle. A complete metabolic response (CMR) was defined by normalisation of the scan appearances, and non-CMR by any residual focal abnormality corresponding to a site of disease on baseline evaluation. FMISO imaging was performed at baseline (within 1 week of the FDG PET study), at 4 weeks and at the end of radiotherapy (week 8). Static high-resolution mode images of the neck were acquired at 2 h after approximately 120 MBq of radiotracer on the same PET scanner as was used for FDG imaging, again in 3-D acquisition mode. On each occasion PET images of the neck were acquired and processed using measured attenuation correction and iterative reconstruction [11] in the treatment position. After co-registration with the baseline FDG PET study as detailed above, tumoural ROIs were applied for semi-quantitative analyses of FMISO uptake in the primary and nodal sites (Fig. 1). Qualitative evaluation and determination of SUVmax were performed blinded to CT results but cognisant of the FDG PET distribution. For qualitative reading, the images were displayed on both a linear grey scale and a square rainbow colour scale. The same scheme was used as for scoring the intensity of uptake as for FDG PET. However, a score of 4 could only be given if the FMISO uptake was qualitatively as intense and as extensive as the corresponding region of FDG uptake. No lesion fulfilled this criterion. The FMISO scan was interpreted as positive if there was greater activity (score of 2 or 3) within the sites of tumoural uptake of FDG than was present in adjacent or mirrored soft tissue sites on the 2-h images when reviewed using a square rainbow colour scale.

Evaluation of local response and clinical follow-up Response was evaluated by examination findings including endoscopy and repeat CT scanning at 12 weeks after completion of treatment (week 19), coinciding with the post-treatment FDG PET

European Journal of Nuclear Medicine and Molecular Imaging Vol. 32, No. 12, December 2005

1386 were counted regardless of cause, and survival times for living patients were censored at the close-out date. Failure-free survival was time to failure defined as either locoregional failure, distant metastases or death without failure, and was censored at the close-out date and date of last follow-up. Time to locoregional failure was defined as time to local or nodal failure. Time to locoregional failure was censored by distant failure, death without preceding locoregional failure, the closeout date and date of last follow-up.

Results Patient characteristics From January 1997 to March 1998, 16 eligible patients were enrolled on this study. They had a mean age of 49±8 years (range 35–69). The primary tumour site was oropharyngeal in 13, oral cavity in two and pyriform fossa in one. There were ten patients with T4 and/or N3 disease representing extremely advanced locoregional disease. There were 15 patients who underwent both FDG and FMISO PET and one who had only serial FDG PET scanning. FDG PET findings at baseline

Fig. 1. Comparison of FDG PET (top) and FMISO PET (middle) demonstrates focal abnormality in relationship to both the primary lesion and a large nodal metastasis. Fusion of the FMISO scan with an isocontour (white line) generated from the FDG PET study (lower panel) after image co-registration demonstrates that the extent of FMISO uptake is less than that of FDG abnormality, consistent with a hypoxic subpopulation of cells. The primary lesion was graded as 4 on FDG and as 2 on FMISO by both readers. The larger, more intense nodal mass was graded as 4 on FDG and as 3 on FMISO. A consensus score of 2 is required for a lesion to be graded as hypoxic for the descriptive purposes of this manuscript. Note that both the extent and the intensity of FDG uptake relative to background tissues were substantially higher on FDG than on FMISO

scans, and by surgical pathology when available. Patients were regularly followed up in a specialist head and neck cancer clinic. In general, patients were followed up for 5 years in this clinic, and further follow-up was then obtained from the patient’s medical record or contact with the family physician.

The PET study demonstrated abnormal FDG uptake at the site of known primary disease in all patients (sensitivity for primary, 100%). In one patient with known oropharyngeal disease, the extent of FDG PET abnormality seemed greater than on CT, with extension into the region of the nasopharynx, and subsequently represented a site of early marginal relapse. The SUVmax values ranged from 5.1 to 23.2, with a mean of 12.1±4.6 (Table 1). Blinded FDG PET reading also demonstrated abnormal uptake at all but one site of known nodal disease identified by conventional staging with CT. One large but hypodense lesion presumed to be an almost completely necrotic lymph node was not seen. In two patients abnormal FDG uptake was apparent in nodes that were not enlarged on CT. These nodes were not verified by biopsy but were incorporated in the radiation treatment volume. In two cases extracervical abnormalities were identified. In one case with uptake adjacent to the dome of the right hemidiaphragm, both high-resolution CT and respiratory-gated magnetic resonance imaging were negative and the patient thus remained on the trial but subsequently relapsed at this site. In the other, abnormality in the lung was presumed to reflect an infective process after correlation with CT, and inclusion in the trial was again deemed to be appropriate. However, subsequently this was proven to be a second primary lung cancer.

Statistics

FMISO PET findings at baseline Overall survival, failure-free survival and time to locoregional failure curves were calculated using the Kaplan-Meier (product-limit) method. The close-out date for survival analyses was August, 2004. The potential follow-up time for each patient was the time from treatment start to the close-out date. For overall survival, all deaths

We have previously reported [8] that 14/15 patients had increased uptake of FMISO involving at least one site of known disease based on clinical reporting. However, on

European Journal of Nuclear Medicine and Molecular Imaging Vol. 32, No. 12, December 2005

1387 Table 1. Patient details and baseline imaging appearances Pt. Stage/ Primary FDG Primary no. primary site SUVmax FMISO SUVmax

Primary Nodal FDG Nodal FMISO Nodal FMISO score SUVmax SUVmax FMISO score

Status at close-out date, follow-up duration (years)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

T3N2A/ oropharynx T4N1/BOT T3N3/tonsil T3N2C/ tonsil T3N1/ oropharynx T2N3/BOT T2N2C/ tonsil T3N3/tonsil T3N3/tonsil T3N2B/ tongue T4N1/BOT T3N2C/ BOT T1N3/BOT T4N2C/ tongue T2N3/pyriform fossa T2N3/tonsil

21.2

3.5

2

9.4

2.7

1

AFFP, 7.5

14.3 18.3 14.3

2.9 3.2 3.1

3 2 2

NV 15.5 7.1

NV 2.3 2.9

NV 2 2

DDD, 4.0 AFFP, 7.5 DSP (lung), 7.4

9.1

3.0

2

4.8

2.7

1

DLR, 4.5

10.6 9.7

3.2 2.4

2 1

12.9 10.1

3.7 2.7

3 1

DSP (lung), 5 LTFU, 6.1

23.2 18.2 5.6

2.3 2.15 1.8

1 2 2

12.0 10.6 2.0

2.5 2.0 1.6

1 3 1

DLR, 0.75 LTFU, 6.1 DLR, 5.7

7.4 8.2

2.5 1.7

3 2

4.2 3.9

1.5 1.7

3 2

DDD, 1 DFFPa , 1

9.1 10.6

ND 1.7

ND 2

5.2 2.8

ND NV

ND NV

DDD, 1.3 DLR, 1

5.1

2.1

1

9.0

2.5

2

DSP (lung), 3.8

7.8

1.6

2

9.0

1.4

2

AFFP, 6.5

BOT base of tongue, AFFP alive free from progression, DFFP died, free from progression, DLR died with local recurrence, DDD died of distant disease without local recurrence, DSP died of a remote second primary (location), LTFU lost to follow-up, ND not done, NV not visualised a This patient remained in complete radiological and clinical response at review 1 month prior to a sudden death of unknown cause

blinded reading by two consensus readers for this report, there were 13 patients with a score of 2 or 3 either in the primary or at a nodal metastatic site. One case was scored as 1 by one reader and 2 by the second and was agreed on consensus to be negative (score 1). All other scores were in agreement. No lesion was graded as 4. Of the 15 known primary tumours visualised on FDG PET, 12 were also qualitatively scored as hypoxic on FMISO PET. The SUVmax for FMISO within the primary tumour ROI defined on the FDG PET study ranged from 1.6 to 3.5, with a mean of 2.5±0.5 (Table 1). The primary lesions not qualitatively scored as positive all had recorded values of SUVmax of less than 2.5 in the corresponding ROI, but below this value some lesions were still graded qualitatively as being positive. Of the 13 patients who had nodal metastases apparent on FDG PET, qualitative evaluation of FMISO was scored as positive in eight. The highest recorded SUVmax for FMISO within any individual nodal region ROI defined on the FDG PET study ranged from 1.4 to 3.7, with a mean of 2.3±0.5. One patient (no. 15) had a positive nodal site

despite lack of visualisation of the primary on FMISO PET. In all cases the SUVmax recorded for FMISO was less than that for FDG at the corresponding site (Table 1). Qualitative PET response All 16 patients had achieved a partial metabolic response on FDG by week 4, and at 12 weeks post-treatment only four patients had failed to achieve a CMR. These four patients all relapsed within 6 months of commencement of treatment. In two cases the progressive disease occurred at distant sites seen on PET but unconfirmed on conventional imaging and there were two cases of early locoregional relapse: one at the edge of the radiation treatment volume (a presumed geographic miss) and only one case related to an area of baseline FMISO uptake (a T4 base of tongue lesion). On qualitative evaluation at week 4, no patient had residual focal accumulation of FMISO apparent at a site of baseline FDG abnormality. As reported in our previous publication [8], unblinded clinical reading revealed a single

European Journal of Nuclear Medicine and Molecular Imaging Vol. 32, No. 12, December 2005

1388

residual focus of abnormal FMISO uptake at 4 weeks. This was in relationship to a large, necrotic nodal mass that was apparent on CT but that had not been associated with definite FDG abnormality on either baseline or follow-up scanning. Using the blinded methodology of this study, this focus could not be graded as positive. The patient involved had resection of the residual neck mass and was found to have a 6-mm focus of residual cancer in a node largely replaced by fibrosis. Clinical efficacy We have previously reported the early clinical results of this trial [8]. In summary, 14/16 patients (88%, 95% CI 62%–98%) achieved a complete response at the primary site on clinical and CT examination by 12 weeks postradiotherapy. In five of these 14 patients, however, persistent neck masses remained at sites of nodal metastasis. Despite negative FDG PET scans, all five patients proceeded to elective neck dissection, with a pathological complete response found in four cases. Hence, 13 out of 16 patients (81%, 95% CI 54–96%) achieved a complete locoregional response following chemo-radiation, and an additional patient was rendered disease-free following neck dissection. Follow-up was extended beyond 5 years in all surviving patients, with a potential median follow-up time of 6.9 years (range 6.1–7.6 years). Two patients were lost to follow-up, although both had been followed up for 6.1 years. Seven patients have failed; three had locoregional failure, one had simultaneous locoregional and distant failure and three had distant failure without any evidence of

local recurrence. Two of the local recurrences occurred very late after treatment (4.5 and 5.7 years). The latter of these, in particular, was strongly suspected to represent a metachronous head and neck cancer rather than a true local relapse. A further three patients have died due to a second lung primary while one died suddenly at 1 year of unknown causes despite a complete response in the primary, a pathological CR in a lymph node mass that had been resected at week 12 and ongoing clinical CR 1 month prior to death. Three patients remained alive and free from progression at last follow-up (6.5, 7.5 and 7.5 years). The 5-year overall survival was 50% (95% CI 27–73%), the 5-year failure-free survival was 44% (95% CI 22–68%) and the 5-year freedom from locoregional failure was 68% (95% CI 38–88%) (Fig. 2).

Discussion Phase I trials traditionally provide only limited information regarding the efficacy of new therapies since initial dosing levels are possibly sub-therapeutic whereas at later dosing levels, toxicity may mask beneficial therapeutic effects. Additionally, patients with advanced disease and poor prognosis are generally selected. In this clinical trial, these inclusion biases may have specifically selected for a subgroup of patients who could uniquely benefit from hypoxia-targeted therapy. The high prevalence of increased FMISO uptake is consistent with the advanced stage of disease in the patients in this trial. A high prevalence of hypoxia would ordinarily predict for an adverse prognosis in such patients [4, 5]. Consequently, the resolution of

Fig. 2. Overall survival, failurefree survival and time to locoregional failure (Kaplan-Meier curve)

European Journal of Nuclear Medicine and Molecular Imaging Vol. 32, No. 12, December 2005

1389

FMISO abnormality with hypoxia-targeting therapy and the excellent local control rates support the rationale for tirapazamine in this setting. Our data further indicate that FMISO PET can provide information that may be useful in selecting patients or disease types that would benefit from hypoxia-targeting therapy. Besides indicating the use of specific hypoxic cell cytotoxins like tirapazamine, the ability to non-invasively detect hypoxia and to monitor its persistence or resolution during treatment would open the way for other rational therapeutic interventions. For example, early surgical salvage might be indicated in patients whose tumours do not show complete resolution of hypoxia. Another possibility would be to use high precision intensity modulated radiation therapy to achieve selective dose escalation to hypoxic tumour sub-regions that can be characterised as being likely to contain radioresistant tumour cells. This concept is based on the idea of defining a “biological target volume” in addition to the conventional anatomical target volumes now used to plan radiotherapy [12]. Of interest, the detection of occult metastatic disease on FDG PET in this series altered the radiation treatment volume in two patients with more extensive nodal disease than expected on conventional staging and predicted subsequent systemic relapse in two patients with disease outside the radiation treatment volume. We have previously reported the phenomenon of increasing likelihood of PETdetected metastases with advancing local stage in patients with lung cancer [13]. Accordingly, FDG PET may be important in excluding patients with distant metastases from radical chemoradiation protocols designed for patients with locoregionally advanced head and neck cancer. There is an emerging understanding of the molecular basis of the aggressive phenotype of hypoxic tumours. This includes evidence of hypoxia-upregulating genes with downstream signalling that increases metastatic potential [14]. Previous studies have validated FMISO as a hypoxiaimaging agent in both animals and humans, with increased retention of FMISO being influenced by oxygenation status and correlated with independent measures of tissue oxygen tension [15]. Generally, increased FMISO has been shown to be associated with polarographic oxygen levels below 10 mmHg, which are thought to represent radiobiologically important hypoxia. Furthermore, recent results have suggested that high FMISO uptake predicts for a poor outcome in head and neck cancer treated with radiotherapy alone [16]. These results were independent of the method used to evaluate FMISO uptake. These methods included modelling of dynamic uptake, tumour–muscle ratio, tumour–mediastinal ratio and SUV calculation. One of the limiting factors in the use of hypoxia imaging is the lack of a standardised method of defining what represents “hypoxia”. There is also likely to be a difference in what constitutes “imageable hypoxia” and the micro-environmental hypoxia that is likely to exist in many tumours [17]. The University of Washington group have previously proposed the use of tumour to blood ratios as a re-

producible method to quantify FMISO PET results [18]. They have defined a tumour to blood ratio of 1.6 as representing hypoxia in a lesion and have then calculated what is effectively a “functional hypoxic tumour volume” based on the number of voxels with a tumour to blood ratio of 1.4 in the lesion ROI defined. While this appears to be a valid approach, it requires a methodologically complex acquisition and processing protocol, with 1 h of summed dynamic frames from 120 to 160 min and multiple blood samples. At the time that our study was commenced, we opted for a simplified clinical protocol acquiring a localised static scan over the neck at 120 min post injection to minimise the inconvenience imposed on patients by the multiple imaging investigations mandated by the study protocol. Since no blood sampling was performed and, because of the limited imaging field of view, no significant blood pool containing structures were available to allow retrospective calculation of tumour to blood ratios in our patients. However, subsequent evaluations have demonstrated that background SUV in blood pool at 2 h is around 0.8–1 and therefore the minimum SUVmax recordings for lesions qualitatively scored as hypoxic in our series (1.6 for primary lesions and 1.4 for nodal lesions) would be consistent with the results of the University of Washington group. We have also subsequently applied our qualitative scoring system to a small cohort of patients from the University of Washington in a blinded manner and compared these with the results of tumour to blood ratios with good concordance (unpublished data, November 2004). There are clear practical advantages with such a qualitative approach as long as it can be reproducibly applied and the results can be demonstrated to be correlated with other known surrogates for tumour hypoxia, and, perhaps most importantly, it can provide independent prognostic information. With regard to reproducibility, there was excellent agreement between two blinded readers, with only one lesion scored discrepantly rated as being hypoxic by blinded evaluation. Larger series will be required to assess the reproducibility of the scoring system proposed and to determine whether it provides prognostic stratification. We have, however, recently presented preliminary data in a randomised setting suggesting that qualitative dichotomisation into hypoxic and non-hypoxic tumours is predictive of outcome in head and neck cancer patients [19]. In our series, there was no SUV cut-off that provided concordant separation into the positive and negative groups defined on qualitative reading. This is probably because qualitative interpretation is determined more by lesion contrast than absolute intensity. While this might suggest that tumour to background activity ratios (TBR) may provide a better concordance with qualitative reading, we found very poor reproducibility of TBR, which appeared to be highly dependent on the background ROI assigned. A recent study including patients with both head and neck cancer and lung cancer found that FMISO uptake both on SUV and TBR was predictive of regional failure [20]. The

European Journal of Nuclear Medicine and Molecular Imaging Vol. 32, No. 12, December 2005

1390

methodology of this study included early dynamic imaging as well as delayed static images from which both SUV and TBR results were recorded. Although the authors found useful information was obtained from the pattern of radiotracer accumulation in the early dynamic phase, possibly reflecting vascularity rather than hypoxia per se, the delayed parameters were independently predictive of outcome. The recorded SUV values in their 13 head and neck cancer patients that were obtained at 2 h and 4 h were closely correlated and had a similar range (1.1–4.3) to that recorded in our series (1.4–3.7). In their series, all head and neck cancer patients with a SUV>2 or a tumour to muscle ratio of >1.6 at 4 h had local treatment failure. In our series (Table 1), 11/15 patients had SUVs >2 at 2 h in their primary. Despite this, only three of these patients had local failure (at 4.5, 0.75 and 1 years). One of these patients had a very high SUV in the primary but was graded as negative on consensus qualitative FMISO scoring. Early relapse occurred at the margin of the radiotherapy field in this patient and was considered to represent a probable geographic miss. One further patient with an SUV>2 died of distant metastatic disease at 4 years despite achieving a durable local complete response, while two others died of a metachronous lung cancers at 3.8 and 4.9 years. The five remaining patients were still alive at 5 years. Thus, despite a high a priori likelihood of hypoxia based on the inclusion criteria and a poor prognosis for the group as a whole, the continued good long-term local control of disease would further validate the positive predictive value of early metabolic response on FMISO PET as an indicator of response in the hypoxic cell fraction to tirapazamine. It is likely that the prevalence of baseline hypoxia will be lower in a less highly selected group of patients with head and neck cancer, and probably also in other cancer types. In these situations FMISO PET may identify patients who might benefit from the addition of tirapazamine or other therapies targeting tumour hypoxia. Conclusion A high prevalence of imaging evidence of tumoural hypoxia and its early resolution during treatment combined with an excellent long-term outcome in a group of patients with poor prognosis disease supports the rationale for using agents such as tirapazamine in locally advanced head and neck cancer. Assuming that the uptake of FMISO observed is indicative of biologically important hypoxia, the results of hypoxia imaging in this population suggest that this technique might aid selection or stratification of patients in trials involving hypoxic-targeting chemotherapeutic agents such as tirapazamine. Acknowledgements. The clinical trial was sponsored by SanofiSynthelabo. We acknowledge the contributions of Dr. John Sachinidis and Dr. Henri Tochon-Danguy of the radiochemistry group, Centre for PET, Austin Hospital, to the synthesis of 18F-FMISO used in this study.

References 1. Pignon JP, Bourhis J, Domenge C, Designe L. Chemotherapy added to locoregional treatment for head and neck squamouscell carcinoma: three meta-analyses of updated individual data. MACH-NC Collaborative Group. Meta-analysis of chemotherapy on head and neck cancer. Lancet 2000;355:949–55 2. Gray LH, Conger AD, Ebert M, Hornesy S, Scott OC. The concentration of oxygen dissolved in tissues at the time of irradiation as a factor in radiotherapy. Br J Radiol 1953;26:638– 48 3. Gatenby RA, Kessler HB, Rosenblum JS, Coia LR, Moldofsky PJ, Hartz WH, et al. Oxygen distribution in squamous cell carcinoma metastases and its relationship to outcome of radiation therapy. Int J Radiat Oncol Biol Phys 1988;14:831–8 4. Nordsmark M, Overgaard M, Overgaard J. Pretreatment oxygenation predicts radiation response in advanced squamous cell carcinoma of the head and neck. Radiother Oncol 1996;41: 31–9 5. Nordsmark M, Overgaard J. A confirmatory prognostic study on oxygenation status and loco-regional control in advanced head and neck squamous cell carcinoma treated by radiation therapy. Radiother Oncol 2000;57:39–43 6. Brizel DM, Sibley GS, Prosnitz LR, Scher RL, Dewhirst MW. Tumor hypoxia adversely affects the prognosis of carcinoma of the head and neck. Int J Radiat Oncol Biol Phys 1997;38:285–9 7. Brown JM. SR 4233 (tirapazamine): a new anticancer drug exploiting hypoxia in solid tumours. Br J Cancer 1993;67:1163– 70 8. Rischin D, Peters L, Hicks R, Hughes P, Fisher R, Hart R, et al. Phase I trial of concurrent tirapazamine, cisplatin, and radiotherapy in patients with advanced head and neck cancer. J Clin Oncol 2001;19:535–42 9. Brown JM, Lemmon MJ. Potentiation by the hypoxic cytotoxin SR 4233 of cell killing produced by fractionated irradiation of mouse tumors. Cancer Res 1990;50:7745–9 10. Dorie MJ, Brown JM. Tumor-specific, schedule-dependent interaction between tirapazamine (SR 4233) and cisplatin. Cancer Res 1993;53:4633–6 11. Benard F, Smith RJ, Hustinx R, Karp JS, Alavi A. Clinical evaluation of processing techniques for attenuation correction with 137Cs in whole-body PET imaging. J Nucl Med 1999;40: 1257–63 12. Ling CC, Humm J, Larson S, Amols H, Fuks Z, Leibel S, et al. Towards multidimensional radiotherapy (MD-CRT): biological imaging and biological conformality. Int J Radiat Oncol Biol Phys. 2000;47(3):551–60 13. MacManus MP, Hicks RJ, Matthews JP, Hogg A, McKenzie AF, Wirth A, et al. High rate of detection of unsuspected distant metastases by PET in apparent stage III non-small-cell lung cancer: implications for radical radiation therapy. Int J Radiat Oncol Biol Phys 2001;50:287–93 14. Pennacchietti S, Michieli P, Galluzzo M, Mazzone M, Giordano S, Comoglio PM. Hypoxia promotes invasive growth by transcriptional activation of the met protooncogene. Cancer Cell 2003;3:347–61 15. Rasey JS, Martin GV, Krohn KA. Quantifying hypoxia with radiolabeled fluoromisonidazole: pre-clinical and clinical studies. In: Machulla H-J, editor. Imaging of hypoxia. Dordrecht: Kluwer Academic Publishers; 1999. p. 85–117 16. Eschmann SM, Paulsen F, Reimold M, Dittmann H, Welz S, Reischl G, et al. Prognostic Impact of hypoxia imaging with 18 F-misonidazole PET in non-small cell lung cancer and head and neck cancer before radiotherapy. J Nucl Med 2005;46:253– 60 17. Evans SM, Koch CJ. Prognostic significance of tumor oxygenation in humans. Cancer Lett 2003;195:1–16

European Journal of Nuclear Medicine and Molecular Imaging Vol. 32, No. 12, December 2005

1391 18. Rasey JS, Koh WJ, Evans ML, Peterson LM, Lewellen TK, Graham MM, et al. Quantifying regional hypoxia in human tumors with positron emission tomography of [18F]fluoromisonidazole: a pretherapy study of 37 patients. Int J Radiat Oncol Biol Phys 1996;36:417–28 19. Hicks R, Rischin D, Fisher R, Binns D, Corry J, Porceddu S, et al. Superiority of tirapazamine-containing chemoradiation in head and neck cancers showing evidence of hypoxia on 18Fmisonidazole (FMISO) PET scanning. Proc Am Soc Clin Oncol 2004;23:491

20. Eschmann S-R, Paulsen F, Reimold M, Dittman H, Welz S, Reischl G, et al. Prognostic impact of hypoxia imaging with 18 F-misonidazole PET in non-small cell lung cancer and head and neck cancer before radiotherapy. J Nucl Med 2005;46:253– 60

European Journal of Nuclear Medicine and Molecular Imaging Vol. 32, No. 12, December 2005