The Clinical Impact of ¹⁸F-FDG PET in Patients with Suspected or Confirmed Recurrence of Colorectal Cancer: A Prospective Study

Patient Population

Between February 15, 1998, and November 13, 1999, 106 consecutive patients who had been referred to our tertiary oncology center for a clinically indicated ¹⁸F-FDG PET scan to evaluate suspected or confirmed regional recurrence of colorectal cancer were prospectively entered into a database. None had confirmed disseminated metastases, and all were potential candidates for aggressive therapy if localized disease were confirmed. In cases of equivocal structural imaging results, patients were included if their referring clinician indicated that aggressive local therapy remained appropriate. The PET facility had been operational for 6 mo before the study began, and the reporting physician had significant PET experience. Patients entered into clinical trials gave written informed consent, and the protocol was approved by the institutional ethics committee.

On the routine clinical request form, the referring specialist oncologists were required to state the indication for the PET scan, the histology of the tumor, the clinical stage, and the results of the most recent structural imaging tests. The referring physicians were also required to state what their proposed management plan would be in view of all available clinical information and investigations up to the time of referral. These investigations routinely included CT and any other imaging or biochemical studies that had previously been undertaken to address specific clinical issues. The treatment options available were not restricted to any single modality so that physicians were able to choose combinations of therapy that they thought appropriate to their individual patient.

Four patients, for whom the physicians had not prospectively entered their management plan before dispatch of the PET result, were excluded. The remaining 102 patients (65 men, 37 women; age range, 28–88 y; mean age, 62 ± 12 y) formed the study population. Prospective management plans were supplied for 96, whereas for 6 the referring clinician refused to commit to a treatment plan without access to the PET information. The original tumor locations were the rectum in 36%, the rectosigmoid colon in 22%, the sigmoid colon in 6%, the descending colon in 4%, the transverse colon in 9%, the ascending colon in 9%, the cecum in 12%, and unspecified in 2%. These patients were studied between 6 wk and 12 y after their initial, potentially curative treatment of colorectal cancer. The median time from treatment to the PET study was 2 y.

Review of Conventional Staging Investigations

Most but not all patients had undergone helical CT with both oral and intravenous contrast media unless contraindicated. Whenever possible, the results of imaging not performed by our institution were evaluated by specialist radiologists at our site for the technical adequacy of the scan and the accuracy of the report. Any imaging studies deemed to be of suboptimal quality were repeated before the PET scan, and the results were provided to the referring clinician. CT was performed within 2 mo of PET in 91 (89%) of 102 patients, and all imaging was performed within 5 mo. The median time between CT and PET was 25 d. In some cases, sonography or MRI had been performed, and its results were used in determining the disease status by conventional evaluation. The results of imaging studies were presented at a weekly multidisciplinary colorectal unit meeting attended by specialist oncologic radiologists, medical and radiation oncologists, and surgeons in determining the management options for patients of this institution and were generally discussed before a PET scan was ordered.

Classification of Referral Indication and Management Plan

From the clinical request provided, the patients’ referral indications were subdivided into the following diagnostic subgroups: limited (solitary or adjacent) metastases amenable to aggressive local therapy (usually surgical resection but including regional chemotherapy, cryoablation, or local radiotherapy), rising carcinoembryonic antigen (CEA) levels with either normal or equivocal structural imaging findings, increasing local symptoms without definite structural abnormalities, known recurrence but extent characterization needed for treatment planning, and new mass or residual mass suggestive of recurrence but unconfirmed by biopsy.

For purposes of analysis, patient management definitions were defined as follows. Active treatment was when any combination of surgery, radiotherapy, or chemotherapy was delivered to patients considered to have recurrence. Supportive treatment was when recurrence was considered proven but no active treatment was given. Observation was when no objective evidence of disease was present and routine surveillance was planned. Treatment planning was when the referring clinicians refused to commit to a treatment plan without access to the PET information.

PET Scan Procedure

All patients were asked to fast for 6 h before the study (21) but were encouraged to drink water. To reduce the confounding effects of changing bladder activity on assessment of the pelvis and to reduce out-of-field scatter in 3-dimensional scanning mode, patients were catheterized when possible and given 10–20 mg furosemide 30 min before imaging. In the later part of the study, patients also received bowel preparation on the day before scanning (22). This was not, however, used in patients with prior colostomy. Imaging was performed using a GE QUEST 300H scanner (UGM Medical Systems, Inc., Philadelphia, PA) at least 1 h after intravenous injection of ¹⁸F-FDG. This scanner is a dedicated 3-dimensional PET scanner that uses NaI detectors (23). Consistent with the restricted count-rate capacity of the NaI detectors on this scanner, patients were administered 70–120 MBq to achieve an estimated singles count rate of <3 million counts per second. Whenever possible, transmission and emission scans were obtained from the neck to the upper femurs to include the lower neck, lung, liver, and adrenal glands in the field of view. Total imaging time was approximately 45 min, and patients were scanned with their arms raised unless unable to tolerate this position. Some patients with claustrophobia were scanned only to the mid thoracic level.

Image datasets were obtained using iterative reconstruction (ordered-subset expectation maximization method) (24). Attenuation correction was performed using previously described methods (25). Datasets with and without measured attenuation correction were reported from the screen using an interactive display program that allows multiple orthogonal images to be displayed simultaneously. Rotating count-rendered images were also reviewed, particularly to clarify the relationship between suspected abnormalities and physiologic radiotracer accumulation in the colon.

PET Scan Interpretation

The PET stage was determined by incorporating the PET findings with all other staging information and was abstracted from the clinical report issued thereafter. In almost all cases, the previous CT findings were directly correlated with the PET findings, but in occasional cases, the images were not available and the CT report was relied on. As a general rule, ¹⁸F-FDG PET abnormalities less intense than physiologic uptake in the liver were not considered to be positive unless there was a small corresponding structural imaging abnormality. This is consistent with the concept of compensation for partial-volume limitations of the PET scanner (26). Significant structural imaging abnormalities did not influence the PET result if there was no associated ¹⁸F-FDG PET abnormality.

Evaluation of Impact of PET Scan and Outcome

All referring physicians were contacted within 1 mo of the scan. Each physician provided information on how patients were managed after their PET scans and when surgical interventions such as laparotomy were performed, operative findings were obtained, and pathology reports were obtained. At least 6 mo and approximately 12 mo after the PET scan report was issued, information on clinical outcome and subsequent imaging or pathologic findings was obtained from the referring clinicians or medical records.

The PET study was classified as true-positive if recurrent disease was confirmed at surgery or on subsequent follow-up, usually including confirmation by structural imaging. The criterion for imaging confirmation was increasing or new abnormalities in the region of PET-defined abnormalities compatible with tumor and sufficient to justify a management change. This criterion did not require biopsy confirmation of every site of PET abnormality. True-positive PET scans were considered to have underestimated the extent of disease if additional sites of involvement were found at subsequent surgery or were confirmed on imaging within l y of the PET scan. A false-positive PET scan was defined as either misclassification of a subsequently proven benign abnormality as malignant or PET abnormalities that showed no progression after at least 2 y of follow-up. PET was considered true-negative if the patient was free of disease progression at least 1 y after the scan. PET was considered false-negative if no abnormality was seen on PET but malignancy was subsequently found at surgery or if progressive abnormalities developed on structural imaging within 1 y of follow-up in patients with durable disease control locally or at the site of subsequent surgery.

Statistics

When appropriate, 95% confidence intervals (CI) calculated using StatXact software (version 4.0.1; CYTEL Software Corp., Cambridge, MA) are reported.

Overall Clinical Impact of ¹⁸F-FDG PET on Management

The impact of PET is shown in Table 1. In the 6 patients for whom the referring clinician was not prepared to commit to a treatment plan without access to the PET information, PET correctly guided management in all patients, with 2 confirmed true-negative studies and 4 validated true-positive results. Of the remaining 96 patients with prospectively defined treatment plans, incremental information provided by ¹⁸F-FDG PET directly changed the management of 54 (56%). Thus, overall, PET directly influenced management decisions in 60 (59%) of 102 patients.

There were 2 patients with positive PET findings that altered management but who died within 3 mo of the PET study without definite confirmation of the incremental findings and 1 other patient who was lost to follow-up. Of the remaining 57 patients, the correct disease status, and therefore appropriate information to guide management change, was validated by clinical follow-up in 52 (91%). These included 7 with true-negative and 45 with true-positive PET results. One patient had a false-positive result relating to a pelvic abscess but was appropriately managed by surgery, whereas in the other 4 patients (7%), the PET result was true-positive but underestimated the extent of disease subsequently confirmed. These included 3 patients with additional small peritoneal metastases (all <1 cm) found at laparotomy and 1 patient with early relapse of extrahepatic disease after resection of a hepatic metastasis. Thus, 49 (98%) of 50 evaluable patients with positive PET findings had confirmation of relapse.

Evaluation of Clinical Indication Subgroups

The patient referral indications and associated changes in management induced by incremental PET findings are presented in Table 1. The largest group consisted of 34 patients with apparently limited metastases thought to be amenable to aggressive local therapy. Substantial changes in management were observed for every diagnostic subgroup.

The most frequent change in management was to vary the active treatment planned. In 26 patients, the PET results directly influenced the changes made, representing 25% (95% CI, 17%–34%) of the entire patient population. Similarly, PET influenced a change from observation to active therapy in 12 patients (12%; 95% CI, 5%–19%) and from active to supportive palliative therapy after showing more extensive disease than previously suspected in 11 patients (11%; 95% CI, 4%–17%). In 5 patients for whom active treatment was changed to observation only, abnormal findings on conventional imaging were thought to be most likely caused by a recurrence, but subsequent PET findings were normal. In 6 further patients, the clinician refused to commit to a treatment plan before availability of the PET result, and PET was therefore deemed to have changed management.

A high impact was found on the delivery of aggressive local therapies, including surgery and local radiotherapy (Table 2). These therapies were correctly prevented by incremental PET findings in 31 patients but initiated in 18 patients. Overall, for 26 (60%) of 43 patients for whom surgical intervention had prospectively been planned, this was abandoned as a direct result of the PET findings. The high impact of PET was particularly evident in the group of patients with apparently limited metastases for whom surgical resection was planned. In this group, only 10 came to operation, including 2 for whom the PET findings of inoperability were confirmed. One patient was recognized at the time of reporting to have false-negative PET findings in relationship to a small (<1 cm) but biopsy-proven hepatic metastasis. This patient underwent limited hepatic resection because there was no evidence of extrahepatic disease on PET, and there was no evidence of progressive extrahepatic disease during more than 12 mo of follow-up. Seven other patients were judged to have potentially operable disease on the basis of concordant CT and PET findings. In 1 of these patients, both CT and PET underestimated the extent of disease found at surgery and curative resection was not attempted, whereas 6 underwent curative surgery and the surgical findings were consistent with the imaging results. Of these, 3 remained free of disease at >1 y, 1 had late recurrence (19 mo), and 2 had early recurrences (6 and 9 mo). All patients with relapsing disease presumably had microscopic disease undetected by CT, ¹⁸F-FDG PET, or diagnostic laparotomy. Of the 19 patients with apparently solitary liver metastases on CT, PET found additional hepatic metastases in 2, extrahepatic disease in 8, and more extensive intrahepatic spread and extrahepatic disease in 3.

In 13 (65%) of 20 patients with a rising CEA level, true-positive incremental PET findings suggested a recurrence site and altered management. In 4 further patients, the PET findings were abnormal but did not alter management: In 1 of these patients, who was described earlier, the findings were misinterpreted through lack of direct correlation with structural imaging; in 3, true-positive PET findings did not induce a management change. Despite the high sensitivity of PET in this patient subgroup, PET underestimated the extent of disease in 4 of these 16 true-positive cases: 2 patients with additional peritoneal metastases (<1 cm) documented at surgery and 2 patients with relapse at additional sites after 4 and 9 mo of further observation. On clinical follow-up, 2 of the 3 PET-negative patients in this subgroup are free from disease at >1.5 y, both with falling CEA levels, but 1 was found to have small-volume peritoneal disease at laparoscopy. In all but 1 patient for whom PET altered management and follow-up was available, the PET result was correct for the relapse status (Table 3), but in several patients detailed above, all imaging techniques underestimated the extent of disease.

Outcomes in Which ¹⁸F-FDG PET Findings Were Inconsistent with Final Management

Of the entire study group, the final management for 18 patients was not consistent with the disease status reported on PET, including 6 patients for whom management was changed from the prospectively assigned treatment plan. The details of these patients are shown in Table 4. In summary, PET was clearly correct for both the presence and the extent of relapse in 9 patients but was inappropriately ignored. In 2 patients, PET correctly identified relapse but underestimated the extent of disease. Nevertheless, inappropriate hepatic resection was avoided in both patients by subsequent evaluation. In another patient, an interpretative false-negative study was recognized by direct correlation of the CT and PET images, emphasizing the limitations of relying on clinical CT reports alone when reporting PET results. In 4 patients, the PET findings were false-negative, but in 2 of these patients the false negativity was prospectively recognized on the basis of knowledge of the presence of a biopsy-confirmed, <l-cm hepatic metastasis on CT. For these 2 patients, correct exclusion of extrahepatic disease allowed surgery to proceed. The other 2 patients had small peritoneal deposits that were missed by PET and CT but were directly visualized by invasive procedures. One patient who was thought to have relapse on PET refused further investigation and has been lost to follow-up, whereas the remaining patient was reported to have probable granulomatous disease as the cause of additional ¹⁸F-FDG PET abnormalities in the mediastinum. These findings were not further evaluated by the treating clinician and therefore did not adversely affect management. On the basis of these findings, the incorporation of PET into clinical decision making leads to appropriate management in most patients. Figures 1 and 2 are 2 representative examples of this.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Both CT and ¹⁸F-FDG PET suggested presence of single hepatic metastasis. At laparotomy, additional small (<1 cm) metastases were found and hepatic resection was abandoned.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

This patient was being considered for resection of an apparently isolated lung metastasis. In addition to lung lesion shown in upper panel, PET found additional multiple liver metastases not apparent on CT, as shown in lower panel. Surgery was avoided, and abdominal metastases were subsequently confirmed by serial imaging.

Effect of CT Timing Relative to PET

Although nearly all patients underwent CT shortly before PET, 5 patients experienced a delay of >3 mo. It is thus possible that these 5 patients may have skewed the overall findings of this study in favor of PET because of the natural progression of the disease rather than increased sensitivity of PET. This subgroup included 3 patients with ongoing symptoms suspected to be caused by recurrence and previously unhelpful structural imaging and for whom the referring clinician elected to use PET as the sole restaging technique. Two had their most recent CT performed 103 and 134 d before a negative PET scan. Both continued with observation without evidence of recurrent disease. A further patient with negative findings on CT performed 154 d before PET was found to have widespread metastatic disease, and adjuvant chemotherapy was therefore stopped. Two patients with apparently isolated metastasis on CT performed 111 and 99 d before PET were found to have disseminated disease. The management of the first was changed from surgery to chemotherapy, and the management of the second was changed from intraarterial chemotherapy to palliative care. Thus, progressive disease between the CT and PET scans could have accounted for the discrepant results in 3 patients, all of whom had their management changed.