| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Mallinckrodt Institute of Radiology St. Louis, MO 63110
Fox Chase Cancer Center Philadelphia, PA 19111
Johns Hopkins University Baltimore, MD 21287
Williams et al. [1] recently reported that positron emission tomography (PET) with FDG had poor sensitivity (24.5%) and specificity (77.3%) for detection of pelvic nodal metastases based on their findings in a retrospective study of 18 women with gynecologic cancer (predominantly cervical cancer; n = 16). By comparison, CT and MR imaging had better sensitivity (48.1% and 53.7%, respectively) and specificity (97% and 90.7%, respectively). We find the report by these investigators to be disconcerting for two principal reasons.
First, it appears that the authors were unaware of prior studies addressing the detection of pelvic nodal metastases by FDG PET in women with cervical cancer [2,3,4]. Rose et al. [2] reported both the sensitivity and specificity of PET to be 100% in a series of 17 women; CT definitively diagnosed adenopathy in only two (22%) of 11 patients with pathologically proven nodal metastases. Sugawara et al. [3] also found higher sensitivity of PET than CT. Grigsby et al. [4] showed that PET detected more pelvic nodal abnormalities than did CT (or lymphangiography); the clinical significance of these abnormalities has been documented in a subsequent study [5], by the finding of a shorter progression-free survival among patients with PET-positive, CT-negative nodal disease than of those considered node negative by both PET and CT. Also, more recently, Reinhardt et al. [6] have documented greater sensitivity and specificity of FDG PET than of MR imaging for detection of pelvic nodal metastases in cervical cancer. Although Williams et al. [1] do cite some of the relevant prior literature pertaining to the detection of nodal metastases by CT and MR imaging, they surprisingly ignored a comprehensive 1997 meta-analysis of this literature by Scheidler et al. [7].
Second, if Williams et al. [1] had cited the earlier literature showing FDG PET to be more accurate for detection of pelvic nodal metastases than they observed, they would have been obligated to explain the apparently discrepant results of their study. Accordingly, the article lacks a thorough discussion of the study's potential limitations. These limitations relate primarily to the technical details of PET imaging as performed in the study by Williams et al. The scanner used (ECAT ART; Siemens-CTI, Knoxville, TN) is a partial-ring bismuth germanate PET scanner operated in the three-dimensional mode. Body images obtained with this scanner tend to be count limited and to have poorer contrast (attributable to the inclusion of more random events) than do images obtained with full-ring scanners operated in the two-dimensional mode. A relatively lower administered dose of FDG (88-222 MBq) was reported by Williams et al. than by other investigators (370-740 MBq). However, imaging times were similar, thus increasing the likelihood that the images analyzed by Williams et al. were count poor. Although this was not stated in the article, the images appear to have been reconstructed by filtered back projection rather than by an iterative method, thus rendering them more subject to artifacts related to focally increased activity in the ureters or bladder [8]. No attenuation correction was used, which means that distortions of the images in the pelvis adjacent to increased activity in the bladder are more likely to have occurred [8]. Surprisingly, no patient preparation was used in five of the 18 patients studied. In the other 13 patients, efforts to decrease urinary tract activity (hydration and intramuscular furosemide) were less vigorous than those used by other investigators (intravenous hydration, intravenous furosemide, and bladder catheterization or irrigation).
In summary, the report by Williams et al. [1] does not reflect the results that have been achieved (and should be achievable by FDG PET with modern full-ring scanners and optimal reconstruction methods) for detection of pelvic nodal metastases in women with cervical carcinoma or other gynecologic neoplasms. To the contrary, the preponderance of the published data indicate that PET is accurate for this purpose and contributes substantially to clinical decision making in patients with these neoplasms.
References
Imperial College Faculty of Medicine London W12 0HS,
England
Addenbrookes Hospital Cambridge CB2 2QQ, England
Siegel, Adler, and Wahl raise some interesting and relevant points. However, we wish to reemphasize that, in contrast to the studies quoted by them, we specifically sought to compare our imaging findings with the gold standard of histopathology of the excised nodes. This meant that the cancers included in our study [1] were at a much earlier stage (International Federation of Gynecology and Obstetrics Ib or lower) than those quoted in other published series, to ensure that our patients underwent primary surgical treatment with lymph node dissection. This selection resulted in a lower prevalence of positive nodes [2] compared with studies published by Rose et al. [3], in which patients were mostly at International Federation of Gynecology and Obstetrics stage IIb and above, and accounts for the lower sensitivity with PET in our study for detecting positive nodes. More important, all our patients had histologic confirmation of tumor presence or absence within the lymph nodes, not merely confirmation of tumor at the primary site. Sugawara et al. [4] confirmed histology of nodes in only one patient, the others being confirmed by clinical follow-up. It was not clear how absence of lymph node involvement at the time of imaging was confirmed in that study. Grigsby et al. [5] did not confirm the presence of nodal metastases on histology because "pelvic and paraaortic nodes are considered to be expected sites of metastases." Also, some of the articles cited are recent publications and antedate our submission [6, 7].
Our study design was substantially different from other published series. In our study, readers were completely unaware of clinical and other imaging information, except for the initial diagnostic assessment, for which clinical information was supplied. These criteria did not apply to other studies [3,4,5,6]. On the contrary, Rose et al. [3] used a negative CT scan as a patient selection criterion. Also, because we did not use ambiguous categories such as "probably normal, probably abnormal" we may have achieved lower specificities with PET than quoted by other groups [4].
Siegel, Adler, and Wahl's second point underlines technical issues that can arise when using partial-ring as opposed to full-ring PET scanners. The scanner used in our study was a fully three-dimensional dedicated PET scanner. Its performance has been described [8]. Because it is fully three-dimensional, the accepted fraction of scatter events is higher for this scanner than for a conventional two-dimensional system, which may impact on contrast. However, effective scatter correction algorithms have been implemented for this scanner [9]. The peak noise equivalent count rate for this camera is greater than 30,000 cycles per second for a 20-cm diameter cylinder at a concentration of 15 kBq/mL. This rate is generally greater than the activity concentration we encounter in our clinical FDG studies (<10 kBq/mL equivalent). It is true that a lower dose is injected for a three-dimensional scan than for a conventional study in two-dimensional mode, but total counts acquired are not significantly different. The difference in dose is offset by the increase in sensitivity gained by operating in three-dimensional mode, in spite of the system's being only a partial ring and not a full ring. The partial-ring scanners make PET imaging available at an affordable cost and provide reasonable accuracy; we optimized our scanning protocols for the scanner capabilities at the time of the study.
The patients in this study were scanned between November 1996 and October 1998. Filtered back-projection was used because iterative reconstruction software was not commercially available to our department at the time of the study. A running debate exists regarding transmission images and the extra time needed to acquire them before obtaining the emission images [10]. At the time of this study, it was not possible to obtain emission—transmission images.
We acknowledge that patient preparation to minimize urinary artefact is important, but we would emphasize that our results in paraaortic lymph nodes, although better than in pelvic nodes, were still disappointing. We explored the use of bladder catheterization both with and without bladder irrigation. Not only did the procedure increase radiation dose to the operator without clearing the problem of residual urine around the catheter tip, but also it added a level of discomfort to the patient, and thus we believed its routine use could not be justified.
The literature on FDG PET imaging of metastatic lymph nodes in the pelvis is still small, and although some studies may be encouraging, we and others [11] regard its use as preliminary. The real value of FDG PET for this application awaits further well-controlled trials. A combination of morphologic and metabolic images may further improve diagnosis [12]. It will be interesting to see what impact on the results will be seen with variations in the PET imaging equipment used.
We thank Siegel, Adler, and Wahl for bringing up these issues for discussion.
References
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
