- Email: [email protected]
The Impact of FDG PET on the Management of Occult Primary Head and Neck Tumours
Clinical Oncology (2003) 15: 461–466 doi:10.1016/j.clon.2003.07.006
Original Article The Impact of FDG PET on the Management of Occult Primary Head and Neck Tumours W. L. Wong*, M. Saunders† *The Paul Strickland Scanner Centre, Mount Vernon Hospital, Northwood, Middlesex; †Marie Curie Research Wing for Oncology, Mount Vernon Hospital, Northwood, Middlesex, U.K. ABSTRACT: Aims: The aim of this study was to investigate the impact of positron-emission tomography (PET) with 18F-labelled fluoro-2deoxy-D-glucose (FDG) in the management of occult primary head and neck tumours. Materials and methods: We reviewed 16 patients with squamous cell carcinoma (SCC) and one patient with undifferentiated carcinoma of cervical lymph nodes (N1-4; N2a-9; N2b-2; N3-2). All patients had full clinical assessment, including examination under anaesthesia (EUA), with biopsy of all suspicious areas and random biopsies of sites likely to harbour an occult primary site. Nine patients underwent magnetic resonance imaging (MRI) of the head and neck, three underwent computed tomography (CT) and five underwent both CT and MRI. None of these studies were able to locate a primary tumour. Patients received 350 MBq FDG intravenously. Emission transmission scans of the extra-cranial head, neck and thorax were obtained using an ECAT Exact 47 at least 60 min after injection. The images were interpreted by the same radiologist experienced in PET, independent of the final outcome. The influence of FDG PET on management was assessed on review of the patients’ notes after treatment or when treatment had been deemed unnecessary. Results: FDG PET suggested a primary site in eight of the 17 patients (tongue base 5; nasopharynx 1; tonsil 1; supraglottis 1). Pathological confirmation was obtained in four patients and one patient died of progressive disease at the primary site. In nine patients, the primary site was not identified on FDG PET. In six of these patients, no primary site was found during follow-up (range 8–36 months; mean 20 months). One patient died before treatment commenced, and there were two histologically confirmed false-negative FDG PET results: one tonsil SCC and one lateral pharyngeal wall SCC. FDG PET affected treatment plans in nine of the 17 (53%) patients in whom a primary site was suggested (altered radiotherapy plan 6; radiotherapy with curative intent to palliative radiotherapy 1; radiotherapy to surgery and post-operative radiotherapy 1), and in one patient where no occult primary was localised (radiotherapy to surgery 1). FDG PET had a sensitivity, specificity, positive and negative predictive value of 62%, 66%, 62% and 62%, respectively. Wong W. L., Saunders M. (2003). Clinical Oncology 15, 461–466 2003 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved. Received: 29 April 2002
Revised: 23 July 2003
Introduction
Carcinomas of the upper-air and food passages are a varied and important group of tumours. Worldwide, head and neck cancer is the sixth most common type of malignancy and, in the U.K., it accounts for about 4% of all cancers [1,2]. Most patients present with disease in one of the major primary sites, with or without nodal involvement. One to two per cent of head and neck cancers, however, present with malignant nodes in the neck without evidence for a primary tumour. These patients undergo a full clinical assessment, laryngoscopy/fibre-optic endoscopy and, if the primary tumour is not found, an examination under anaesthesia (EUA), with biopsy of any suspicious lesions and areas likely to harbour a primary tumour. Computed Author for correspondence: Dr Wai Lup Wong, Consultant Radiologist, The Paul Strickland Scanner Centre, Mount Vernon Hospital, Northwood, Middlesex, HA6 2RN, U.K. Tel.: +44-0-1923844751; Fax: +44-0-1923-844600; E-mail: [email protected] 0936-6555/03/080461+6 $30.00/0
Accepted: 24 July 2003
tomography (CT) and more recently magnetic resonance imaging (MRI), complement clinical assessment in the search for the primary malignancy. Despite these extensive investigations, the primary tumour is not found in some patients, and they present a difficult clinical problem for treatment. It is standard practice at our hospital to irradiate the whole of the head and neck area, including nasopharynx, oropharynx, hypopharynx and larynx to a microscopic dose of 44–50 Gy in 4–5 weeks, and a high dose of between 60–66 Gy is given to the nodal disease. This wide-field irradiation is associated with significant acute morbidity and late morbidity, particularly in terms of xerostomia. Identification of the primary site can allow for more directed treatment, with a higher dose to the known primary and a lower dose to the surrounding normal tissues, thus reducing morbidity. 18F-labelled fluoro-2-deoxy-D-glucose (FDG) positron-emission tomography (PET) has been shown in several studies to be of use in the assessment of patients with occult head
2003 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
462
CLINICAL ONCOLOGY
and neck malignancies. However, the ability of FDG PET to detect a primary tumour when no primary site has been identified after thorough clinical examination, including EUA with multiple biopsies and radiological assessment, is unclear. Furthermore, no studies to date have addressed the impact of FDG PET on the management of the occult primary head and neck tumour in the U.K. To assess the value of FDG PET in the head and neck area, we conducted a retrospective audit of patients that had been scanned in our department.
Materials and Methods
From January 1999 to September 2000, 21 patients with a diagnosis of occult head and neck primary were referred to the Scanner Centre. Seventeen of these patients had been given a full clinical assessment. This included CT or MRI, and an EUA with biopsy of all suspicious areas and multiple random biopsies of sites likely to harbour an occult primary tumour, including the nasopharynx, oropharynx, hypopharynx and larynx. These revealed no primary site, and these patients formed the population of our study, of which 12 were men and five were women. Ages ranged from 50–87 years (mean 61 years). Sixteen patients had biopsyproven squamous cell carcinoma (SCC), and one patient had undifferentiated carcinoma in the neck nodes. Eleven patients presented with right-sided nodes, and six with left-sided nodes. One patient presented with level 1 nodes, 11 with level 2 nodes and five with level 2 and 3 nodes. Clinical staging were N1 (4 patients), N2a (9 patients), N2b (2 patients) and N3 (2 patients). Nine patients underwent head and neck MRI, only three patients had CT because of claustrophobia, and five patients had both MRI and CT. Nine MRI scans and five CTs were performed at our own centre. The CT and MRI scans were reviewed before FDG PET to confirm that no primary site was visible (WLW). All patients then underwent FDG PET. Where anatomical localisation of the suspected primary site was unclear, CT/MRI was registered to FDG PET. PET FDG studies were obtained using an ECAT Exact 47 (CTI/Siemens, Knoxville, TN) scanner. All patients fasted for at least 12 h and were normoglycaemic at the time of scanning. Scanning was performed at least 60 min after 350 MBq of FDG was administered intravenously. Attenuation-corrected data were obtained through the extra-cranial head, neck and chest. Both emission and post-injection transmission data were acquired for 15 min per bed position (90 min of scanning for three bed positions), with emission scanning commenced 60 min after injection. ‘Measured’ attenuation correction was applied after correcting the transmission data for emission contamination. The images were interpreted by the same radiologist experienced in PET imaging (WLW), independent of the final outcome. The FDG PET was always read with the assistance of either image registration or at least
comparing FDG PET and anatomical imaging side by side. All patients scanned at our institution underwent a spiral CT examination (Siemens plus 4, images obtained in an axial plane with a collimation of 3 mm, reconstruction thickness, 4 mm table feed after 150 ml non-ionic contrast). MRI examination was performed on a 1.5T whole-body imager (Siemens, Siemens Medical Systems, Erlangen, Germany) Coronal STIR, T1W sequences were complimented by axial T1W and turbo-spin echo T2 sequences. The medical records of all patients were reviewed, and the results of the PET FDG scan were correlated with findings at clinical assessment and surgery. The influence of FDG PET on management was assessed on review of the patients’ notes after treatment or when treatment had been deemed unnecessary (MS). For FDG PET scans that were positive for tumour, but not confirmed histologically, a true positive was scored when there was progression of disease at the primary site to death, otherwise a false-positive result was recorded. For FDG PET examinations that were negative for tumour, and a confirmation biopsy was not performed, a true negative was scored when the relapse-free survival was achieved for a minimum of 8 months.
Results
The clinical characteristics, imaging findings and impact of FDG PET on the management of these patients is summarised in Table 1. FDG PET suggested a primary tumour in eight of the 17 patients; five in the base of the tongue, one in each of the nasopharynx, tonsil and supraglottis. It was not possible to go back and confirm the diagnosis histologically in all patients but, in four patients in whom this was attempted, all gave positive histology. One other patient died of progressive disease at the primary site, giving a true positive rate of 62% (5/8). The patient with a nasopharyngeal primary was not re-biopsied owing to poor general condition, and the patient died 1 month after treatment. Two patients, in whom base-of-tongue tumours were diagnosed on PET FDG, received radiotherapy with modified fields, excluding the nasopharynx and giving a full tumourcidal dose to the oropharynx; both these patients are alive and well at 24 and 28 months after radiotherapy, and only time will tell whether local recurrence will occur and the FDG PET be proved correct. In six patients, the site of primary tumour was not identified nor found subsequently. In five patients, there was no change in treatment policy, and radiotherapy was given to the whole head and neck area, as described earlier. In one patient, there was a change of treatment policy from radiotherapy to surgery to the neck without post-operative radiotherapy. All patients are alive without evidence of disease from 8–30 months after treatment, giving a true negative rate of 66% (6/9).
FDG PET AND OCCULT PRIMARY HEAD AND NECK TUMOURS
463
Table 1 – Patient characteristics, imaging findings and management impact of FDG Primary site seen
Primary site/confirmation of primary site
MRI/CT
Clinical outcome/follow-up period
FDG PET
Not found Not found Not found BOT/seen on repeat indirect endoscopy after FDG BOT/progression of disease at BOT despite treatment Not biopsied
CT 0 MRI 0 CT 0 MRI 0 CT 0 MRI 0 MRI 0
TN TN TN TP BOT
NED NED NED NED
MRI 0 MRI 0
TP BOT mediastinal node FP NP
Not biopsied Tonsil/tonsillectomy
MRI 0 MRI 0
FP BOT TP tonsil
Not found Not found Not found Lateral pharyngeal wall/resection of primary site (9 mm carcinoma) BOT/re-biopsy of area suspicious on FDG Not confirmed
CT 0 MRI 0 CT 0 MRI 0 CT 0 MRI 0 MRI 0
TN TN TN FN
Dead; progressive disease at primary site despite RT Died 1 month after completing treatment from cirrhosis NED 28 months NED 18 months but lung cancer 9 months after head and neck treatment NED 36 months NED 8 months NED10 months NED 24 months
CT 0; no MRI because of claustrophobia CT 0; no MRI because of claustrophobia CT 0 MRI 0
TP BOT
NED 4 months
Yes
FP BOT
NED24 months
Yes
TP Supraglottis + lung metastases* FN 0
NED 12 months
Yes
NED; developed Gullain Barrie syndrome 1 month after completiing treatment 5 months Died just before treatment commencement
No
Supra-glottis/re-biopsy of area suspicious on FDG Tonsil/tonsillectomy Not biopsied
CT 0 no MRI because of claustrophobia MRI 0
0 0 0 0
FN 0 bilateral neck nodes
22 15 24 14
months months months months
Management impact No No No Yes Yes Yes Yes Yes Yes No No No
No
*No lung metastases were identified on chest X-ray 1 week before FDG. BOT, base of tongue; FDG, 18F-labelled fluoro-2-deoxy-D-glucose; FN, false negative; FP, false positive; NED, no evidence of disease in the extra-cranial head and neck; NP, nasopharynx; RT, radiation therapy; TN, true negative; TP, true positive; 0, not seen.
In one patient, FDG PET did not show the primary site but showed unsuspected contralateral neck nodes. The patient died before confirmation of the bilateral neck-nodal disease. In a further two patients, PET FDG did not show a cancer, but a primary tumour was subsequently diagnosed: in one patient, a 9 mm carcinoma on the lateral pharyngeal wall, which was probably too small to be diagnosed by FDG PET and, in another, a carcinoma of the tonsil, which was confirmed histologically at tonsillectomy. False-negative rate was 33% (3/9). In this study, PET increased the yield of primary tumour by 29% (5/17) and affected the management in 53% (9/17). The sensitivity, specificity, positive and negative predicted values in this series of PET FDG were 62% (5/8), 66% (6/9), 62% (5/8) and 66% (6/9), respectively. Discussion
The ability of FDG PET to detect disease consistently and reliably at the primary site in head and neck cancer patients has encouraged its use for the assessment of the occult primary cancer [3–6]. Rege et al. [5] studied four patients with unknown primary lesions, and found that FDG PET located the primary base of tongue cancers in
two of the four patients and magnetic resonance in none of the four patients. Since then, various studies have specifically investigated the role of FDG PET in this area [7–16]. The corner stone for investigating this group of patients is with EUA, including multiple biopsies, CT and MRI; FDG PET can only be justified if it can detect more primary sites after these investigations. In three studies, which included between 13 and 27 patients, FDG PET detected between 21–30% more primary sites compared with conventional work-up [10,14,15]. Some head and neck cancer units, including ours, favour the use of MRI for detecting submucosal head and neck cancer. A review of the literature yielded only one study in which MRI was clearly used in a significant number of patients with metastatic SCC cervical lymph nodes with unknown primary tumours. It included 20 patients of whom seven had CT, nine had MRI, two had CT plus MRI and two had no imaging [20]. FDG PET correctly identified 7/10 (70%) of histologically proven primary sites [16]. In contrast, among the nine CT scans, two primary sites were identified and, of the 11 MRI scans performed, four primary sites were identified. All the lesions seen on MRI, and one of the two lesions seen on CT were demonstrated on FDG PET. These results
464
CLINICAL ONCOLOGY
suggest that FDG PET can detect more occult primaries after MRI as well as CT. It is unclear, however, how many of these patients had EUA and biopsies before FDG PET. In our series, FDG PET increased the yield of primary tumours by 8/17 (47%) after conventional assessment, of which 5/17 (29%) were confirmed by subsequent clinical course or pathology. All patients had EUA with multiple biopsies before FDG PET, and most had MRI alone or CT plus MRI rather than CT alone. We report a false-positive rate, specificity and positive-predictive value (PPV) of 37%, 66%, and 62%, respectively, because lesions on FDG PET not confirmed by histology or progressive disease at the primary site identified by FDG PET was considered as falsely positive. A previous study had an apparent false-positive result in 46% (6/13) of patients [12]. In this study, and also in other studies, a negative biopsy was reported as a false-positive result. It is possible that the biopsy was in error and the specificity and true PPV of the technique may be underestimated. In our study, all patients, including the three patients in whom FDG PET showed a primary site that was not subsequently confirmed, were treated on the basis of the FDG PET result. No other primary head and neck tumour developed during subsequent follow-up in any of the patients. This observation supports the speculation that FDG PET may be more specific than the current figures indicate. Alternatively, the false-positive results could have been due to physiological uptake of FDG. Symmetrical lowgrade uptake is consistently seen in lymphoid tissue within Waldeyer’s ring [17]. We have also occasionally observed asymmetric uptake, which can be prominent within lymphoid tissue, and includes the post-nasal space and tonsils in patients with no head and neck malignancy. Further clarification may be available with long-term follow-up of larger groups of patients. Few histologically verified false-positive results were identified in the literature review: 8 (5%) histologically verified false-positive results in a cummulative total of 177 patients [7–16]. We found no definite false-positive results, assuming that to be so, because a biopsy was not performed and local recurrence did not occur after radiotherapy. Inflammatory pulmonary lesions, including granulomas and abscess formation, can result in false-positive FDG PET results [11,14]. Abnormal lung FDG uptake did not contribute to lesions that mimicked malignancy in our study. It may partly reflect the lower prevalence of granulomas in U.K. patients compared with North American patients. Unlike previous studies, FDG uptake, due to dental caries and benign thyroid and salivary gland nodules, did not cause a diagnostic dilemma, partly because it is now recognised that these lesions can cause intense FDG uptake [8,10]. Recent laryngoscopic biopsy was reported in one patient in a study of 17 patients as a cause of false-positive results [11]. However, there have been no other such reports, and was not a cause of increased FDG uptake in this study. We found that tonsillectomy,
on the other hand, can result in potential confusion, not on the side of surgery but on the contralateral side, where normal tonsilar uptake may mimic a tonsillar lesion. False-negative results are unusual with FDG PET. In studies where false-negative results were reported, there was a cumulative total of over 146 patients with occult head and neck primary tumours; only six (4.1%) histologically confirmed false-negative results were identified: base of tongue carcinoma in three, tonsillar carcinoma in two and nasopharyngeal cancer in one [10–12,14–16]. The ability to detect cancers less than 1 cm across with FDG PET is as yet unclear. Using attenuation-corrected scanning, FDG PET showed a 4 mm base of tongue small cell carcinoma (SCC) in one study and demonstrated an 8 mm piriform sinus SCC in another [11,15]. Using the whole-body method, without attenuation correction, FDG PET detected a 5 mm laryngeal SCC and a 3 mm plasmacytoma of the tongue, but failed to show a 4 mm tongue small cell carcinoma (SCC) [10]. We failed with attenuation correction scanning, even in retrospect, to show a 7 mm pharyngeal wall primary squamous cell carcinoma. Most studies report that, in most patients with an occult primary tumour after staging, which includes FDG PET, no head and neck primary tumour will emerge after treatment during follow-up [7–15]. Our results are concordant with this observation; no such patient developed a head and neck cancer during subsequent follow-up of up to 36 months with a mean of 20 months. Studies have shown that FDG PET can further influence the therapeutic plan in between 21% and 55% of patients [7,8,13–15]. In the study by Jungehulsing, et al. [15], FDG PET affected management in two patients in whom FDG PET was unable to identify a primary site, as well as those patients in whom the primary was located. FDG PET identified unsuspected metastases to the clavicle in one patient and mediastinal nodes in another, which resulted in extension of radiation fields [15]. In another study [11], FDG PET showed an unsuspected synchronous oesophageal carcinoma. In our study, FDG PET influenced treatment plan in 53% of patients, including all the patients in whom a primary site was found. In two of these patients, FDG PET suggested distant metastases. This led to palliative radiotherapy in the patient with pulmonary metastases detected on FDG PET but not seen on chest X-ray. It influenced treatment plan in one patient where no occult primary was localised. In this patient, FDG PET, in concordance with clinical assessment and imaging, suggested TX N2a MO disease, and treatment was changed from empiric radiotherapy to radical neck dissection. The role of radical radiotherapy to the mucosa in the head and neck area remains controversial. Results from non-randomised studies indicate that such wide-field radiotherapy reduces the incidence of appearance of the primary tumour in the head and neck area, but at
FDG PET AND OCCULT PRIMARY HEAD AND NECK TUMOURS
465
0 D OLJ Q D Q W6 & & Q H F N Q R G H
5 H H [ D P LQ H WK H + 1 LQ F OX G LQ J OD U\ Q J R V F R S \ IOH [ LE OH H Q G R V F R S \
0 05 5 & & 7 7 3 R V V LE OH S ULP D U\V LWH
1 R R E Y LR X V S ULP D U\ V LWH
) )' '* * 3 3 ( 7 7 ( 8 $ D Q G E LR S V LH V Fig. 1 – Proposed algorithm for investigating occult head and neck injuries.
considerable morbidity. The European Organisation for Research and Treatment in Cancer (EORTC) is launching a randomised-controlled trial comparing neck irradiation only to wide-field radiation in individuals with an unknown primary. The results of this trial should inform the management of these patients, including the role of FDG PET in this area. The ultimate value of FDG PET is to focus EUA on areas most likely to yield positive histology. The head and neck surgeon can potentially use FDG PET to obtain a larger sample from suspicious areas, thereby increasing the diagnostic yield compared with speculative biopsies. Several studies suggest that biopsies carried out after FDG PET can improve the number of occult primaries detected compared with endoscopic examination with speculative biopsies [7,15]. Registered FDG PET to CT/MRI can further improve anatomical localisation [18]. We found it particularly useful in the oro-pharyx, where it helped distinguish between lesions in the tongue base and tonsil. In our study, positive results were obtained in both patients re-biopsied using information from registered FDG PET. It may have contributed to the more accurate localisation of lesions and hence greater biopsy yield compared with FDG PET without image registration. This study has several limitations. It is a selective retrospective review with all its inherent problems. The MRI and CT scans were performed at various institutions, using varying techniques and interpreted by various observers. FDG PET was assessed by one observer. At the time of original reporting of the FDG PET, all cross-sectional imaging was reviewed before
FDG PET analysis by one observer experienced in head and neck imaging to confirm that the examinations were adequate and that there was no primary site. Interpretation of FDG PET by one observer is considered by some to be less than ideal. We are, however, of the opinion that careful interpretation of the scan, with detailed correlation to cross-sectional imaging by one observer experienced in head and neck PET, is a valid alternative to multiple readings by observers with less head and neck PET experience. At our hospital, all patients with occult primary head and neck tumours after full clinical assessment, laryngoscopy/fibre-optic endoscopy and CT/MRI have FDG PET. FDG PET is performed before EUA and biopsy, as we feel that this investigation algorithm provides the best opportunity for detecting the occult primary while minimising misleading results (Fig. 1).
Conclusion
In conclusion, our results show that FDG PET is a worthwhile technique for assessing patients with occult head and neck primary tumours, and can complement current techniques in this group of patients.
References 1 Grau C, Johansen LV, Jacobson J, et al. Cervical lymph node metastases from unknown primary tumours. Results from a national survey by the Danish Society for Head and Neck Oncology. Radiother Oncol 2000;55:121–129.
466
CLINICAL ONCOLOGY
2 Nieder C, Gregoire V, Kian Ang K. Cervical lymph node metastases from occult squamous cell carcinoma. Int J Radiat Oncol Biol Phys 2001;50:727–733. 3 Wong WL, Saunders M. Role of PET FDG in the management of head and neck squamous cell cancer. Clin Oncol 1998;10:361–366. 4 Wong WL, Chevretton E, McGurk M, et al. A prospective study of FDG PET imaging for the assessment of head and neck squamous cell carcinoma. Clin Otolaryngol 1977;22:209–214. 5 Rege S, Maass A, Chaiken L, et al. Use of positron emission tomography with fluorodeoxyglucose in patients with extracranial head and neck cancers. Cancer 1994;73:3047–3048. 6 Wong WL, Chevretton E, McGurk M, et al. PET-FDG imaging in the clinical evaluation of head and neck cancer. J R Soc Med 1995; 88:469–473. 7 Mukherji SK, Drane WE, Mancuso AA, et al. Occult primary tumours of the head and neck: detection with 2-[F-18] fluore-2-dglucose SPECT. Radiology 1996;199:761–766. 8 Stokkel MPM, Terhaard CH, Hordijk GK, et al. The detection of unknown primary tumours in patients with cervical metastases by dual-head positron emission tomography. Oral Oncol 1999; 35:390–394. 9 Schipper JH, Chhrader M, Arweiler D, et al. Die Positronenemissonstomo-graphie zur Primatumoursuche bei Halslyphknotenmeta – stasen mit unbekanntem Primartumor [in German]. HNO 1996;44:254–257. 10 Braams JW, Pruim J, Kole AC, et al. Detection of unknown primary head and neck tumors by positron emission tomography. Int J Oral Maxillofac Surg 1997;26:112–115.
11 Assar OS, Fischbein NJ, Caputo GR, et al. Metastatic head and neck cancer: role and usefulness of FDG PET in locating occult primary tumours. Radiology 1999;210:177–181. 12 Greven KM, Keyes JW, Williams DW III, et al. Occult primary tumours of the head and neck. Am Cancer Soc 1999;86:1. 13 Bohuslavizki KH, Klutmann S, Sonnemann U, et al. F-18 FDG PET for detection of occult primary tumor in patients with lymphatic metastases of the neck region [in German]. Laryngorhinootologie 1999;78:445–449. 14 Safa A, Tran LM, Rege S, et al. The role of positron emission tomography in occult primary head and neck cancers. Cancer J Sci Am 1999;5:205–207. 15 Jungehulsing M, Schneidhauer K, Damm M, et al. 2 (18 F)-fluoro2-deoxy-D-glucose positron emission tomography is a sensitive tool for the detection of occult primary cancer (cancer of unknown syndrome) with head and neck lymph node manifestation. Otolaryngol Head Neck Surg 2000;123:294–301. 16 Matthew M, Hanasono MD, Larisa D, et al. Uses and limitations of FDG positron emission tomography in patients with head and neck cancer. Laryngoscope 1990;109:880–885. 17 Goerres GW, von Schulthness GK, Hany TF. PET and PET CTY of the head and neck: FDG uptake in normal anatomy, benign lesions, and in changes resulting from treatment. AJR Am J Roentgenol 2002;17:1337–1343. 18 Wong WL, Hussain K, Chevretton E, et al. Validation and clinical application of computer combined CT and PET with FDG head and neck images. Am J Surg 1996;172:628–632.
Original Article The Impact of FDG PET on the Management of Occult Primary Head and Neck Tumours W. L. Wong*, M. Saunders† *The Paul Strickland Scanner Centre, Mount Vernon Hospital, Northwood, Middlesex; †Marie Curie Research Wing for Oncology, Mount Vernon Hospital, Northwood, Middlesex, U.K. ABSTRACT: Aims: The aim of this study was to investigate the impact of positron-emission tomography (PET) with 18F-labelled fluoro-2deoxy-D-glucose (FDG) in the management of occult primary head and neck tumours. Materials and methods: We reviewed 16 patients with squamous cell carcinoma (SCC) and one patient with undifferentiated carcinoma of cervical lymph nodes (N1-4; N2a-9; N2b-2; N3-2). All patients had full clinical assessment, including examination under anaesthesia (EUA), with biopsy of all suspicious areas and random biopsies of sites likely to harbour an occult primary site. Nine patients underwent magnetic resonance imaging (MRI) of the head and neck, three underwent computed tomography (CT) and five underwent both CT and MRI. None of these studies were able to locate a primary tumour. Patients received 350 MBq FDG intravenously. Emission transmission scans of the extra-cranial head, neck and thorax were obtained using an ECAT Exact 47 at least 60 min after injection. The images were interpreted by the same radiologist experienced in PET, independent of the final outcome. The influence of FDG PET on management was assessed on review of the patients’ notes after treatment or when treatment had been deemed unnecessary. Results: FDG PET suggested a primary site in eight of the 17 patients (tongue base 5; nasopharynx 1; tonsil 1; supraglottis 1). Pathological confirmation was obtained in four patients and one patient died of progressive disease at the primary site. In nine patients, the primary site was not identified on FDG PET. In six of these patients, no primary site was found during follow-up (range 8–36 months; mean 20 months). One patient died before treatment commenced, and there were two histologically confirmed false-negative FDG PET results: one tonsil SCC and one lateral pharyngeal wall SCC. FDG PET affected treatment plans in nine of the 17 (53%) patients in whom a primary site was suggested (altered radiotherapy plan 6; radiotherapy with curative intent to palliative radiotherapy 1; radiotherapy to surgery and post-operative radiotherapy 1), and in one patient where no occult primary was localised (radiotherapy to surgery 1). FDG PET had a sensitivity, specificity, positive and negative predictive value of 62%, 66%, 62% and 62%, respectively. Wong W. L., Saunders M. (2003). Clinical Oncology 15, 461–466 2003 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved. Received: 29 April 2002
Revised: 23 July 2003
Introduction
Carcinomas of the upper-air and food passages are a varied and important group of tumours. Worldwide, head and neck cancer is the sixth most common type of malignancy and, in the U.K., it accounts for about 4% of all cancers [1,2]. Most patients present with disease in one of the major primary sites, with or without nodal involvement. One to two per cent of head and neck cancers, however, present with malignant nodes in the neck without evidence for a primary tumour. These patients undergo a full clinical assessment, laryngoscopy/fibre-optic endoscopy and, if the primary tumour is not found, an examination under anaesthesia (EUA), with biopsy of any suspicious lesions and areas likely to harbour a primary tumour. Computed Author for correspondence: Dr Wai Lup Wong, Consultant Radiologist, The Paul Strickland Scanner Centre, Mount Vernon Hospital, Northwood, Middlesex, HA6 2RN, U.K. Tel.: +44-0-1923844751; Fax: +44-0-1923-844600; E-mail: [email protected] 0936-6555/03/080461+6 $30.00/0
Accepted: 24 July 2003
tomography (CT) and more recently magnetic resonance imaging (MRI), complement clinical assessment in the search for the primary malignancy. Despite these extensive investigations, the primary tumour is not found in some patients, and they present a difficult clinical problem for treatment. It is standard practice at our hospital to irradiate the whole of the head and neck area, including nasopharynx, oropharynx, hypopharynx and larynx to a microscopic dose of 44–50 Gy in 4–5 weeks, and a high dose of between 60–66 Gy is given to the nodal disease. This wide-field irradiation is associated with significant acute morbidity and late morbidity, particularly in terms of xerostomia. Identification of the primary site can allow for more directed treatment, with a higher dose to the known primary and a lower dose to the surrounding normal tissues, thus reducing morbidity. 18F-labelled fluoro-2-deoxy-D-glucose (FDG) positron-emission tomography (PET) has been shown in several studies to be of use in the assessment of patients with occult head
2003 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
462
CLINICAL ONCOLOGY
and neck malignancies. However, the ability of FDG PET to detect a primary tumour when no primary site has been identified after thorough clinical examination, including EUA with multiple biopsies and radiological assessment, is unclear. Furthermore, no studies to date have addressed the impact of FDG PET on the management of the occult primary head and neck tumour in the U.K. To assess the value of FDG PET in the head and neck area, we conducted a retrospective audit of patients that had been scanned in our department.
Materials and Methods
From January 1999 to September 2000, 21 patients with a diagnosis of occult head and neck primary were referred to the Scanner Centre. Seventeen of these patients had been given a full clinical assessment. This included CT or MRI, and an EUA with biopsy of all suspicious areas and multiple random biopsies of sites likely to harbour an occult primary tumour, including the nasopharynx, oropharynx, hypopharynx and larynx. These revealed no primary site, and these patients formed the population of our study, of which 12 were men and five were women. Ages ranged from 50–87 years (mean 61 years). Sixteen patients had biopsyproven squamous cell carcinoma (SCC), and one patient had undifferentiated carcinoma in the neck nodes. Eleven patients presented with right-sided nodes, and six with left-sided nodes. One patient presented with level 1 nodes, 11 with level 2 nodes and five with level 2 and 3 nodes. Clinical staging were N1 (4 patients), N2a (9 patients), N2b (2 patients) and N3 (2 patients). Nine patients underwent head and neck MRI, only three patients had CT because of claustrophobia, and five patients had both MRI and CT. Nine MRI scans and five CTs were performed at our own centre. The CT and MRI scans were reviewed before FDG PET to confirm that no primary site was visible (WLW). All patients then underwent FDG PET. Where anatomical localisation of the suspected primary site was unclear, CT/MRI was registered to FDG PET. PET FDG studies were obtained using an ECAT Exact 47 (CTI/Siemens, Knoxville, TN) scanner. All patients fasted for at least 12 h and were normoglycaemic at the time of scanning. Scanning was performed at least 60 min after 350 MBq of FDG was administered intravenously. Attenuation-corrected data were obtained through the extra-cranial head, neck and chest. Both emission and post-injection transmission data were acquired for 15 min per bed position (90 min of scanning for three bed positions), with emission scanning commenced 60 min after injection. ‘Measured’ attenuation correction was applied after correcting the transmission data for emission contamination. The images were interpreted by the same radiologist experienced in PET imaging (WLW), independent of the final outcome. The FDG PET was always read with the assistance of either image registration or at least
comparing FDG PET and anatomical imaging side by side. All patients scanned at our institution underwent a spiral CT examination (Siemens plus 4, images obtained in an axial plane with a collimation of 3 mm, reconstruction thickness, 4 mm table feed after 150 ml non-ionic contrast). MRI examination was performed on a 1.5T whole-body imager (Siemens, Siemens Medical Systems, Erlangen, Germany) Coronal STIR, T1W sequences were complimented by axial T1W and turbo-spin echo T2 sequences. The medical records of all patients were reviewed, and the results of the PET FDG scan were correlated with findings at clinical assessment and surgery. The influence of FDG PET on management was assessed on review of the patients’ notes after treatment or when treatment had been deemed unnecessary (MS). For FDG PET scans that were positive for tumour, but not confirmed histologically, a true positive was scored when there was progression of disease at the primary site to death, otherwise a false-positive result was recorded. For FDG PET examinations that were negative for tumour, and a confirmation biopsy was not performed, a true negative was scored when the relapse-free survival was achieved for a minimum of 8 months.
Results
The clinical characteristics, imaging findings and impact of FDG PET on the management of these patients is summarised in Table 1. FDG PET suggested a primary tumour in eight of the 17 patients; five in the base of the tongue, one in each of the nasopharynx, tonsil and supraglottis. It was not possible to go back and confirm the diagnosis histologically in all patients but, in four patients in whom this was attempted, all gave positive histology. One other patient died of progressive disease at the primary site, giving a true positive rate of 62% (5/8). The patient with a nasopharyngeal primary was not re-biopsied owing to poor general condition, and the patient died 1 month after treatment. Two patients, in whom base-of-tongue tumours were diagnosed on PET FDG, received radiotherapy with modified fields, excluding the nasopharynx and giving a full tumourcidal dose to the oropharynx; both these patients are alive and well at 24 and 28 months after radiotherapy, and only time will tell whether local recurrence will occur and the FDG PET be proved correct. In six patients, the site of primary tumour was not identified nor found subsequently. In five patients, there was no change in treatment policy, and radiotherapy was given to the whole head and neck area, as described earlier. In one patient, there was a change of treatment policy from radiotherapy to surgery to the neck without post-operative radiotherapy. All patients are alive without evidence of disease from 8–30 months after treatment, giving a true negative rate of 66% (6/9).
FDG PET AND OCCULT PRIMARY HEAD AND NECK TUMOURS
463
Table 1 – Patient characteristics, imaging findings and management impact of FDG Primary site seen
Primary site/confirmation of primary site
MRI/CT
Clinical outcome/follow-up period
FDG PET
Not found Not found Not found BOT/seen on repeat indirect endoscopy after FDG BOT/progression of disease at BOT despite treatment Not biopsied
CT 0 MRI 0 CT 0 MRI 0 CT 0 MRI 0 MRI 0
TN TN TN TP BOT
NED NED NED NED
MRI 0 MRI 0
TP BOT mediastinal node FP NP
Not biopsied Tonsil/tonsillectomy
MRI 0 MRI 0
FP BOT TP tonsil
Not found Not found Not found Lateral pharyngeal wall/resection of primary site (9 mm carcinoma) BOT/re-biopsy of area suspicious on FDG Not confirmed
CT 0 MRI 0 CT 0 MRI 0 CT 0 MRI 0 MRI 0
TN TN TN FN
Dead; progressive disease at primary site despite RT Died 1 month after completing treatment from cirrhosis NED 28 months NED 18 months but lung cancer 9 months after head and neck treatment NED 36 months NED 8 months NED10 months NED 24 months
CT 0; no MRI because of claustrophobia CT 0; no MRI because of claustrophobia CT 0 MRI 0
TP BOT
NED 4 months
Yes
FP BOT
NED24 months
Yes
TP Supraglottis + lung metastases* FN 0
NED 12 months
Yes
NED; developed Gullain Barrie syndrome 1 month after completiing treatment 5 months Died just before treatment commencement
No
Supra-glottis/re-biopsy of area suspicious on FDG Tonsil/tonsillectomy Not biopsied
CT 0 no MRI because of claustrophobia MRI 0
0 0 0 0
FN 0 bilateral neck nodes
22 15 24 14
months months months months
Management impact No No No Yes Yes Yes Yes Yes Yes No No No
No
*No lung metastases were identified on chest X-ray 1 week before FDG. BOT, base of tongue; FDG, 18F-labelled fluoro-2-deoxy-D-glucose; FN, false negative; FP, false positive; NED, no evidence of disease in the extra-cranial head and neck; NP, nasopharynx; RT, radiation therapy; TN, true negative; TP, true positive; 0, not seen.
In one patient, FDG PET did not show the primary site but showed unsuspected contralateral neck nodes. The patient died before confirmation of the bilateral neck-nodal disease. In a further two patients, PET FDG did not show a cancer, but a primary tumour was subsequently diagnosed: in one patient, a 9 mm carcinoma on the lateral pharyngeal wall, which was probably too small to be diagnosed by FDG PET and, in another, a carcinoma of the tonsil, which was confirmed histologically at tonsillectomy. False-negative rate was 33% (3/9). In this study, PET increased the yield of primary tumour by 29% (5/17) and affected the management in 53% (9/17). The sensitivity, specificity, positive and negative predicted values in this series of PET FDG were 62% (5/8), 66% (6/9), 62% (5/8) and 66% (6/9), respectively. Discussion
The ability of FDG PET to detect disease consistently and reliably at the primary site in head and neck cancer patients has encouraged its use for the assessment of the occult primary cancer [3–6]. Rege et al. [5] studied four patients with unknown primary lesions, and found that FDG PET located the primary base of tongue cancers in
two of the four patients and magnetic resonance in none of the four patients. Since then, various studies have specifically investigated the role of FDG PET in this area [7–16]. The corner stone for investigating this group of patients is with EUA, including multiple biopsies, CT and MRI; FDG PET can only be justified if it can detect more primary sites after these investigations. In three studies, which included between 13 and 27 patients, FDG PET detected between 21–30% more primary sites compared with conventional work-up [10,14,15]. Some head and neck cancer units, including ours, favour the use of MRI for detecting submucosal head and neck cancer. A review of the literature yielded only one study in which MRI was clearly used in a significant number of patients with metastatic SCC cervical lymph nodes with unknown primary tumours. It included 20 patients of whom seven had CT, nine had MRI, two had CT plus MRI and two had no imaging [20]. FDG PET correctly identified 7/10 (70%) of histologically proven primary sites [16]. In contrast, among the nine CT scans, two primary sites were identified and, of the 11 MRI scans performed, four primary sites were identified. All the lesions seen on MRI, and one of the two lesions seen on CT were demonstrated on FDG PET. These results
464
CLINICAL ONCOLOGY
suggest that FDG PET can detect more occult primaries after MRI as well as CT. It is unclear, however, how many of these patients had EUA and biopsies before FDG PET. In our series, FDG PET increased the yield of primary tumours by 8/17 (47%) after conventional assessment, of which 5/17 (29%) were confirmed by subsequent clinical course or pathology. All patients had EUA with multiple biopsies before FDG PET, and most had MRI alone or CT plus MRI rather than CT alone. We report a false-positive rate, specificity and positive-predictive value (PPV) of 37%, 66%, and 62%, respectively, because lesions on FDG PET not confirmed by histology or progressive disease at the primary site identified by FDG PET was considered as falsely positive. A previous study had an apparent false-positive result in 46% (6/13) of patients [12]. In this study, and also in other studies, a negative biopsy was reported as a false-positive result. It is possible that the biopsy was in error and the specificity and true PPV of the technique may be underestimated. In our study, all patients, including the three patients in whom FDG PET showed a primary site that was not subsequently confirmed, were treated on the basis of the FDG PET result. No other primary head and neck tumour developed during subsequent follow-up in any of the patients. This observation supports the speculation that FDG PET may be more specific than the current figures indicate. Alternatively, the false-positive results could have been due to physiological uptake of FDG. Symmetrical lowgrade uptake is consistently seen in lymphoid tissue within Waldeyer’s ring [17]. We have also occasionally observed asymmetric uptake, which can be prominent within lymphoid tissue, and includes the post-nasal space and tonsils in patients with no head and neck malignancy. Further clarification may be available with long-term follow-up of larger groups of patients. Few histologically verified false-positive results were identified in the literature review: 8 (5%) histologically verified false-positive results in a cummulative total of 177 patients [7–16]. We found no definite false-positive results, assuming that to be so, because a biopsy was not performed and local recurrence did not occur after radiotherapy. Inflammatory pulmonary lesions, including granulomas and abscess formation, can result in false-positive FDG PET results [11,14]. Abnormal lung FDG uptake did not contribute to lesions that mimicked malignancy in our study. It may partly reflect the lower prevalence of granulomas in U.K. patients compared with North American patients. Unlike previous studies, FDG uptake, due to dental caries and benign thyroid and salivary gland nodules, did not cause a diagnostic dilemma, partly because it is now recognised that these lesions can cause intense FDG uptake [8,10]. Recent laryngoscopic biopsy was reported in one patient in a study of 17 patients as a cause of false-positive results [11]. However, there have been no other such reports, and was not a cause of increased FDG uptake in this study. We found that tonsillectomy,
on the other hand, can result in potential confusion, not on the side of surgery but on the contralateral side, where normal tonsilar uptake may mimic a tonsillar lesion. False-negative results are unusual with FDG PET. In studies where false-negative results were reported, there was a cumulative total of over 146 patients with occult head and neck primary tumours; only six (4.1%) histologically confirmed false-negative results were identified: base of tongue carcinoma in three, tonsillar carcinoma in two and nasopharyngeal cancer in one [10–12,14–16]. The ability to detect cancers less than 1 cm across with FDG PET is as yet unclear. Using attenuation-corrected scanning, FDG PET showed a 4 mm base of tongue small cell carcinoma (SCC) in one study and demonstrated an 8 mm piriform sinus SCC in another [11,15]. Using the whole-body method, without attenuation correction, FDG PET detected a 5 mm laryngeal SCC and a 3 mm plasmacytoma of the tongue, but failed to show a 4 mm tongue small cell carcinoma (SCC) [10]. We failed with attenuation correction scanning, even in retrospect, to show a 7 mm pharyngeal wall primary squamous cell carcinoma. Most studies report that, in most patients with an occult primary tumour after staging, which includes FDG PET, no head and neck primary tumour will emerge after treatment during follow-up [7–15]. Our results are concordant with this observation; no such patient developed a head and neck cancer during subsequent follow-up of up to 36 months with a mean of 20 months. Studies have shown that FDG PET can further influence the therapeutic plan in between 21% and 55% of patients [7,8,13–15]. In the study by Jungehulsing, et al. [15], FDG PET affected management in two patients in whom FDG PET was unable to identify a primary site, as well as those patients in whom the primary was located. FDG PET identified unsuspected metastases to the clavicle in one patient and mediastinal nodes in another, which resulted in extension of radiation fields [15]. In another study [11], FDG PET showed an unsuspected synchronous oesophageal carcinoma. In our study, FDG PET influenced treatment plan in 53% of patients, including all the patients in whom a primary site was found. In two of these patients, FDG PET suggested distant metastases. This led to palliative radiotherapy in the patient with pulmonary metastases detected on FDG PET but not seen on chest X-ray. It influenced treatment plan in one patient where no occult primary was localised. In this patient, FDG PET, in concordance with clinical assessment and imaging, suggested TX N2a MO disease, and treatment was changed from empiric radiotherapy to radical neck dissection. The role of radical radiotherapy to the mucosa in the head and neck area remains controversial. Results from non-randomised studies indicate that such wide-field radiotherapy reduces the incidence of appearance of the primary tumour in the head and neck area, but at
FDG PET AND OCCULT PRIMARY HEAD AND NECK TUMOURS
465
0 D OLJ Q D Q W6 & & Q H F N Q R G H
5 H H [ D P LQ H WK H + 1 LQ F OX G LQ J OD U\ Q J R V F R S \ IOH [ LE OH H Q G R V F R S \
0 05 5 & & 7 7 3 R V V LE OH S ULP D U\V LWH
1 R R E Y LR X V S ULP D U\ V LWH
) )' '* * 3 3 ( 7 7 ( 8 $ D Q G E LR S V LH V Fig. 1 – Proposed algorithm for investigating occult head and neck injuries.
considerable morbidity. The European Organisation for Research and Treatment in Cancer (EORTC) is launching a randomised-controlled trial comparing neck irradiation only to wide-field radiation in individuals with an unknown primary. The results of this trial should inform the management of these patients, including the role of FDG PET in this area. The ultimate value of FDG PET is to focus EUA on areas most likely to yield positive histology. The head and neck surgeon can potentially use FDG PET to obtain a larger sample from suspicious areas, thereby increasing the diagnostic yield compared with speculative biopsies. Several studies suggest that biopsies carried out after FDG PET can improve the number of occult primaries detected compared with endoscopic examination with speculative biopsies [7,15]. Registered FDG PET to CT/MRI can further improve anatomical localisation [18]. We found it particularly useful in the oro-pharyx, where it helped distinguish between lesions in the tongue base and tonsil. In our study, positive results were obtained in both patients re-biopsied using information from registered FDG PET. It may have contributed to the more accurate localisation of lesions and hence greater biopsy yield compared with FDG PET without image registration. This study has several limitations. It is a selective retrospective review with all its inherent problems. The MRI and CT scans were performed at various institutions, using varying techniques and interpreted by various observers. FDG PET was assessed by one observer. At the time of original reporting of the FDG PET, all cross-sectional imaging was reviewed before
FDG PET analysis by one observer experienced in head and neck imaging to confirm that the examinations were adequate and that there was no primary site. Interpretation of FDG PET by one observer is considered by some to be less than ideal. We are, however, of the opinion that careful interpretation of the scan, with detailed correlation to cross-sectional imaging by one observer experienced in head and neck PET, is a valid alternative to multiple readings by observers with less head and neck PET experience. At our hospital, all patients with occult primary head and neck tumours after full clinical assessment, laryngoscopy/fibre-optic endoscopy and CT/MRI have FDG PET. FDG PET is performed before EUA and biopsy, as we feel that this investigation algorithm provides the best opportunity for detecting the occult primary while minimising misleading results (Fig. 1).
Conclusion
In conclusion, our results show that FDG PET is a worthwhile technique for assessing patients with occult head and neck primary tumours, and can complement current techniques in this group of patients.
References 1 Grau C, Johansen LV, Jacobson J, et al. Cervical lymph node metastases from unknown primary tumours. Results from a national survey by the Danish Society for Head and Neck Oncology. Radiother Oncol 2000;55:121–129.
466
CLINICAL ONCOLOGY
2 Nieder C, Gregoire V, Kian Ang K. Cervical lymph node metastases from occult squamous cell carcinoma. Int J Radiat Oncol Biol Phys 2001;50:727–733. 3 Wong WL, Saunders M. Role of PET FDG in the management of head and neck squamous cell cancer. Clin Oncol 1998;10:361–366. 4 Wong WL, Chevretton E, McGurk M, et al. A prospective study of FDG PET imaging for the assessment of head and neck squamous cell carcinoma. Clin Otolaryngol 1977;22:209–214. 5 Rege S, Maass A, Chaiken L, et al. Use of positron emission tomography with fluorodeoxyglucose in patients with extracranial head and neck cancers. Cancer 1994;73:3047–3048. 6 Wong WL, Chevretton E, McGurk M, et al. PET-FDG imaging in the clinical evaluation of head and neck cancer. J R Soc Med 1995; 88:469–473. 7 Mukherji SK, Drane WE, Mancuso AA, et al. Occult primary tumours of the head and neck: detection with 2-[F-18] fluore-2-dglucose SPECT. Radiology 1996;199:761–766. 8 Stokkel MPM, Terhaard CH, Hordijk GK, et al. The detection of unknown primary tumours in patients with cervical metastases by dual-head positron emission tomography. Oral Oncol 1999; 35:390–394. 9 Schipper JH, Chhrader M, Arweiler D, et al. Die Positronenemissonstomo-graphie zur Primatumoursuche bei Halslyphknotenmeta – stasen mit unbekanntem Primartumor [in German]. HNO 1996;44:254–257. 10 Braams JW, Pruim J, Kole AC, et al. Detection of unknown primary head and neck tumors by positron emission tomography. Int J Oral Maxillofac Surg 1997;26:112–115.
11 Assar OS, Fischbein NJ, Caputo GR, et al. Metastatic head and neck cancer: role and usefulness of FDG PET in locating occult primary tumours. Radiology 1999;210:177–181. 12 Greven KM, Keyes JW, Williams DW III, et al. Occult primary tumours of the head and neck. Am Cancer Soc 1999;86:1. 13 Bohuslavizki KH, Klutmann S, Sonnemann U, et al. F-18 FDG PET for detection of occult primary tumor in patients with lymphatic metastases of the neck region [in German]. Laryngorhinootologie 1999;78:445–449. 14 Safa A, Tran LM, Rege S, et al. The role of positron emission tomography in occult primary head and neck cancers. Cancer J Sci Am 1999;5:205–207. 15 Jungehulsing M, Schneidhauer K, Damm M, et al. 2 (18 F)-fluoro2-deoxy-D-glucose positron emission tomography is a sensitive tool for the detection of occult primary cancer (cancer of unknown syndrome) with head and neck lymph node manifestation. Otolaryngol Head Neck Surg 2000;123:294–301. 16 Matthew M, Hanasono MD, Larisa D, et al. Uses and limitations of FDG positron emission tomography in patients with head and neck cancer. Laryngoscope 1990;109:880–885. 17 Goerres GW, von Schulthness GK, Hany TF. PET and PET CTY of the head and neck: FDG uptake in normal anatomy, benign lesions, and in changes resulting from treatment. AJR Am J Roentgenol 2002;17:1337–1343. 18 Wong WL, Hussain K, Chevretton E, et al. Validation and clinical application of computer combined CT and PET with FDG head and neck images. Am J Surg 1996;172:628–632.