Oncologic positron emission tomography: a surgical perspective

At one time, positron emission tomography (PET) was used only in academic research. Now it is increasingly being used in clinical practice. This article reviews the background of PET in surgical oncology, its approved indications, and its uses in the diagnosis and management of different types of cancer.

BACKGROUND OF PET

In the vast majority of current clinical applications, PET uses a glucose analog tracer, ¹⁸F-fluorodeoxyglucose (FDG), which has a half-life of 110 minutes. Because of this, PET scanners previously had to be located relatively closely to a cyclotron, the type of accelerator utilized for ¹⁸F production. In recent years, local and regional distribution centers have increased the availability of FDG in most major cities. The principle of using FDG in PET lies in the usefulness of identifying metabolically active cells (Figure 1). Just as chemotherapy utilizes metabolic analogues to treat cancer by exploiting cancer cells' increased metabolism and nucleic acid synthesis, FDG tends to accumulate in cells that are consuming the most glucose. While quantitative analysis is possible, qualitative standardized uptake value analysis of localized consumption is most commonly used for clinical interpretation.

Figure 1.

A normal PET scan. Notice the organs with high glucose consumption, such as the brain and kidneys. The bladder fills as FDG is eliminated. The liver and heart are somewhat less intense. Notice the uptake in the laryngeal region on this patient. This can easily represent the movement of the vocal cords as his arytenoids consume glucose.

INDICATIONS FOR PET

In 1998, Medicare first began to approve reimbursement for PET for several oncologic conditions. In general, Medicare will cover PET for diagnosis, staging, and restaging of a solitary pulmonary nodule, non—small cell lung cancer, melanoma, head and neck cancer, colorectal cancer, esophageal cancer, and lymphoma. Medicare also began to cover PET for breast cancer (staging and restaging of locoregional disease or distant metastases, and treatment follow-up) in October 2002. In addition, Medicare will approve PET costs for the following nononcologic conditions: myocardial perfusion (utilizing a different tracer, Rubidium-82, an analog of potassium), myocardial viability, and refractory epilepsy. Although Medicare will not cover PET costs for sarcoma or for primary brain, pancreatic, hepatocellular, testicular, ovarian, and cervical cancer, PET should be considered for these and other tumors under special circumstances.

As specific conditions are considered with regard to PET scanning, several themes emerge. First, as a functional scan, PET assesses metabolic changes that are usually detectable earlier than anatomic changes. Thus, if an anatomic image like computed tomography (CT) does not show a pancreatic mass but there is high clinical suspicion, PET may be appropriate. Second, PET helps to classify a radiographic abnormality as benign or malignant; the solitary pulmonary nodule is the classic example. When anatomic studies are equivocal, such as when postsurgical scars or radiation necrosis must be distinguished from viable tumor, PET often provides such differentiation. Finally, serial PET examinations can be used to monitor the efficacy of a given treatment regimen. The results may facilitate a change in treatment plan by providing early physiologic evidence showing whether the initial therapy is working. This helps to avoid wasted time, morbidity, and expense. Physicians can then offer the patient more effective and efficient treatment.

Overall, PET has been shown to change treatment plans in many patients, including 45% of the colorectal cancer patients treated at Baylor University Medical Center (1). However, this statistic does not translate into 45% fewer operations. Although occult, distant metastases usually preclude curative resection, PET may show that a patient who was deemed to have disseminated disease is a candidate for curative resection. In the future, even tumors considered inoperable, such as small-cell lung cancer, usually a nonsurgical disease at the time of diagnosis, may eventually prove to be amenable to surgery in some cases if PET can confirm very early, localized disease with sufficient confidence. More importantly, the use of PET to diagnose limited metastases or recurrences earlier than anatomic imaging modalities is likely to improve the surgical cure rate of such patients.

LUNG CANCER

Solitary pulmonary nodule

A wealth of information has been published on the use of PET to assess the solitary pulmonary nodule (Figure 2). Seven studies from 1990 to 1998 show that PET ranges in sensitivity from 83% to 100% in detecting malignancy, with specificity of 63% to 90% (2–8) (Table 1). The largest study to date, by Lowe et al in 1997, evaluated 197 patients and showed that PET has a sensitivity of 96% and specificity of 77% (7). In most hands, the sensitivity and specificity approximate 95% and 85%, respectively. Due to the high glucose utilization of macrophages, numerous inflammatory processes such as granulomatous diseases, sarcoidosis, and aspergillosis can lead to false-positive results.

Figure 2.

A chest CT reveals an abnormal mass (arrow). (b) The PET scan confirms a metabolically active process (arrow), but the mediastinum is free of disease. The patient went directly to surgery, where the mass was found to be a resectable adenocarcinoma of the lung.

Table 1.

Sensitivity and specificity of positron emission tomography in assessing the solitary pulmonary nodule^*

Researcher	n	Sensitivity (%)	Specificity (%)
Kubota et al (1990)	32	83	90
Duhaylongsod et al (1995)	67	97	81
Bury et al (1996)	50	100	88
Knight et al (1996)	48	100	63
Gupta et al (1996)	61	93	88
Lowe et al (1997)	197	96	77
Lowe et al (1998)	90	92	90
Total	555	83–100	63–90

*Compiled from references 2–8.

Non-small cell lung cancer

If a diagnosis of non-small cell lung cancer is made, PET can help surgeons determine which patients would benefit from resection. PET may be especially helpful when traditional anatomic imaging, such as CT or sonogram, is equivocal (Figure 3).

Figure 3.

PET scan demonstrating not only tumor and associated mediastinal metastases but also distant metastases.

How good is PET at staging the mediastinum? In 339 patients in 8 different studies over the past decade, the sensitivity of PET ranged from 75% to 100%, while the sensitivity of CT ranged only from 43% to 87%; the specificity of PET ranged from 81 % to 100%, whereas that of CT was lower, between 44% and 94% (9–16).

Based on an average sensitivity and specificity of 90%, if non-small cell lung cancer at stage N2 has a historical prevalence of 30%, the negative predictive value of a negative PET scan of the mediastinum is 95%. This suggests that no further evaluation is required. Conversely, with the same assumptions for a positive PET scan, the positive predictive value falls to 79%. Because of the lower predictive value, mediastinoscopy should still be performed. More recent studies have shown unexpected distant metastases in 13% to 17% of non-small cell lung cancer patients; thus, PET results upstaged these patients who were not thought to have mediastinal nodes or distant disease (17–19). Correct staging potentially avoids a noncurative thoracotomy.

In patients who have undergone treatment for lung cancer, PET can help measure tumor response and assess for tumor recurrence (Figure 4).

Figure 4.

CT and (b) PET scans of a patient who had radiation therapy for lung cancer. The CT is indeterminate. On the PET scan, the posterior rim is avid and is probably viable tumor.

MacManus et al studied 153 patients with tumors that were deemed unresectable by conventional imaging (20). PET imaging showed that, in fact, 6 of the patients (4%) had tumors that were resectable; two thirds (66%) needed external beam irradiation, and 30% needed palliative treatment. After 2 years of follow-up, PET-assigned staging (P = 0.0041) correlated significantly better with survival than did conventional staging (P = 0.02).

Increasingly, surgeons are beginning to appreciate PET scans as another tool to deal with lung cancer—for the workup of a solitary nodule, for staging before surgery, or for postoperative surveillance.

MELANOMA

The benefits of PET imaging in melanoma are more subtle. Several early PET studies suggested staging “high-risk” melanomas—those >1.5 mm thick—with PET. Data supporting the use of PET in patients with high-risk melanoma tout its sensitivity, which ranges from 74% to 100%, and its specificity, which ranges from 83% to 98% (21–26). There are few treatment and management options for melanoma, but patients who might otherwise have isolated limb perfusion or resection of a solitary metastasis may require a different approach if PET demonstrates distant subclinical metastases (Figure 5).

Figure 5.

PET scan from a patient with melanoma of the leg who has had a popliteal recurrence. Although PET may be accurate in melanoma, its clinical utility in initial staging is limited.

The sentinel lymph node biopsy has removed much of the morbidity of staging intermediate-thickness melanomas; therefore, although PET may be accurate, its clinical utility in initial staging of intermediate-thickness melanoma is rather limited. PET currently should not be considered an alternative to sentinel node biopsy in the staging of regional lymph node basins due to the higher accuracy of directed sampling and evaluation with modern pathologic techniques. In a thick melanoma (>4.0 mm), if physical examination and conventional imaging reveal no metastasis, PET may help determine if systemic chemotherapy is needed, although the effect of PET on survival has not been studied.

HEAD AND NECK CANCER

Nonmelanoma cancers of the head and neck present a variety of problems for which PET scans may be helpful. For staging of head and neck cancer, PET can help identify both locoregional and distant metastases, especially when CT scans are indeterminate. During follow-up visits, PET can monitor the results of treatment. For example, if a patient opted for irradiation but PET demonstrated poor tumor response, the patient's treatment could be altered and surgery performed earlier, when it may be more successful.

Synchronous primary lesion

One phenomenon widely observed in head and neck cancer is that of the synchronous primary lesion. PET is particularly well suited for workup of the second primary lesion. A recent study by Stokkel et al found a synchronous primary lesion in 9 of 54 patients (17%) by using PET (27). For staging, PET was 96% sensitive and 90% specific; this was superior to either CT or combined ultrasound/fine-needle aspiration.

Unknown primary lesion

In the case of an unknown primary lesion, PET may not change the outcome dramatically, but it may guide treatment. Advanced head and neck cancer may respond well to aggressive treatment, while a bronchogenic carcinoma metastatic to the neck would require a different approach. A study by Bohuslavizki et al looked at 53 patients with unknown primary lesions (28). Despite inconclusive conventional imaging, PET was able to identify the true primary site in 20 patients, including one patient with breast cancer and another with cecal cancer (Figure 6).

Figure 6.

Another example from Baylor University Medical Center. One patient had a right cervical node that was found to be squamous cell carcinoma. Thorough evaluation with physical exam, CTs of the neck and chest, and panendoscopy revealed no primary site. The primary site was identified and treated only after PET localized the area of uptake to the vallecula on the left.

Thyroid cancer

PET can also be used to follow up on treatment for papillary and follicular thyroid cancer. In 2001, Frilling et al studied 24 patients with differentiated thyroid cancer and a rising thyroglobulin level but a negative ¹³¹I scan (29). PET was 95% sensitive in detecting recurrences and distant metastases but was only 25% specific. However, PET affected management in 9 of 24 patients (38%). Though there are little hard data, a generally accepted phenomenon is that, as a tumor dedifferentiates, its ability to concentrate ¹³¹ I decreases, while its glucose metabolism increases.

Nevertheless, for medullary and anaplastic thyroid cancers, published results for PET have been less consistent. PET sensitivity and specificity were 78% and 79%, respectively (30). Only magnetic resonance imaging had higher sensitivity at 82%. Both metaiodobenzylguanidine (MIBG) and ^11lIn-Octreoscan studies, however, had higher specificity than PET, at 100% and 92%, respectively. A case report of insular cancer showed no uptake by PET (31). In general, optimal PET imaging of endocrine and neuroendocrine tumors probably awaits approval of positron-emitting tracers other than FDG, such as ¹⁸F-dopa or labeled somatostatin-receptor-binding peptides.

COLORECTAL CANCER

The clearest indication for PET in colorectal cancer is a rising carcinoembryonic antigen level with an inconclusive CT scan. However, almost half of all patients with recurrences will not demonstrate an elevated carcinoembryonic antigen level. Patients frequently benefit from PET if posttreatment conventional imaging shows an indeterminate lesion. Finally, other patients may be candidates for resection of metastases, and PET can help assure the surgeon that the patient may be cured by resection or ablation of isolated metastasis (Figure 7).

Figure 7.

Scans of a patient with colon cancer, (a) CT scan showing a metastasis (arrow) and a benign cyst (arrowhead), which were confirmed by sonography. (b) PET shows the metastasis (arrow), but not the benign cyst, and multiple metastases (arrowheads) not seen on the CT scan.

Many reports have evaluated the effectiveness of PET in diagnosing recurrent colorectal cancer (1, 32–36) (Table 2). At the North Texas Clinical PET Institute, PET was consistently more sensitive than CT in the regions of the liver, pelvis, and abdomen (1). Findings on the 52 patients examined in that report correlated well with the overall literature, which shows an overall PET sensitivity ranging from 85% to 93%. Specificity of PET was consistently better than that of CT, although it was more variable among different centers. PET Institute data show >93% specificity for all regions. With this information, PET scans actually changed the management of 35% to 56% of the patients, including 45% of the patients at the North Texas Clinical PET Institute. A recent prospective trial in Australia found that PET changed patient management in 59% of cases (37).

Table 2.

Sensitivity and specificity of positron emission tomography and computed tomography in recurrent colorectal cancer^*

		Liver				Pelvis				Abdomen
		Sensitivity (%)		Specificity (%)		Sensitivity (%)		Specificity (%)		Sensitivity (%)		Specificity (%)
Researcher	n	PET	CT	PET	CT	PET	CT	PET	CT	PET	CT	PET	CT
Schiepers et al (1995)	76	94	85	100	99	93	60	97	72	—	—	—	—
Delbeke et al (1997)	52	91	81	96	60	100	74	50	40	—	—	—	—
Ogunbiyi et al (1997)	58	95	74	100	85	91	52	100	80	—	—	—	—
Valk et al (1999)	155	95	84	100	95	97	68	96	90	85	50	100	99
Hooker et al (2000)	52	100	79	100	93	86	70	96	93	86	57	93	93
Arulampalam et al (2001)	42	100	45	100	100	100	86	86	100	93	73	75	58

*Compiled from references 1, 32–36.

PET indicates positron emission tomography; CT, computed tomography.

What does the future hold for PET in colorectal cancer? A group in Ohio has reported using a modified handheld gamma counter during surgery to guide the surgeon to the tumor, though the short half-life and expense of FDG make this technique impractical at present (38). In addition, adenomatous polyps have been shown to be PET avid, especially those >13 mm (39). Finally, PET has been shown to surpass intraoperative sonography in the detection and management of colorectal cancer (40).

ESOPHAGEAL CANCER

For esophageal cancer, PET is used during staging to improve patient selection, thus avoiding radical surgery that would not be curative. The Kaplan-Meier plot in Figure 8 demonstrates how survival differed in 100 consecutive patients when based only on PET findings of local vs distant disease. This study helps to validate PET as an independent staging tool (41).

Figure 8.

Kaplan-Meier plot demonstrating how survival differed in 100 consecutive patients when based on PET findings alone. The solid line identifies patients with localized disease by PET; the dashed line, the outcome of those patients found to have distant metastasis after up to 36 months of follow-up. Reprinted from reference 41 with permission of Elsevier Science, Inc.

Additional support comes from an article by Choi et al (42), who found PET to consistently outperform CT at nodal staging and identification of metastases, with sensitivity of 75% vs 18%, specificity of 99% vs 97%, and accuracy of 86% vs 78%, respectively. PET even surpassed endoscopic ultrasound and CT at nodal staging, with rates of 83%, 58%, and 60%, respectively.

PET is limited in esophageal evaluation, however, because esophagitis may lead to false-positive scans. To date, PET has not been shown to identify high-grade dysplasia or carcinoma in situ (43). Thus, patients with these conditions will still require upper endoscopy.

BREAST CANCER

Numerous studies have applied PET to the detection of breast cancer (44–50). Mammography has high sensitivity, but its specificity is <30%. PET has also shown sensitivity ranging from 80% to 100% and specificity ranging from 86% to 100%. PET may miss well-differentiated cancers such as tubular cancers or ductal carcinoma in situ because of their relatively normal metabolism, while inflammatory lesions may lead to false-positive results. Small, subcentimeter lesions may lie below the resolution of PET. On the other hand, patients in whom the value of mammography might be limited may benefit from PET detection of breast cancer. Young women who have dense breasts or women who have undergone breast augmentation are the ideal candidates.

A recent report by Raylman et al evaluated the effectiveness of PET in guiding percutaneous needle biopsy to tumor regions with the highest activity when imaging had revealed diffuse abnormalities (51). The potential value of PET lies in reducing sampling errors during needle biopsy and thus improving the diagnostic accuracy of the procedure.

In staging breast cancer, axillary dissection remains the gold standard. However, PET has performed well. In over 278 patients, the sensitivity of PET varied from 57% to 100%, while specificity ranged from 75% to 100% (44–46, 48, 52–54).

Breast cancer is another process in which the standardized uptake value may reflect tumor metabolism and ultimate histologic grading. Smith and colleagues used this grading information from serial PET scans to identify responders to a given treatment and to change treatment modalities accordingly (55).

In addition to treatment follow-up, PET's largest role in breast cancer currently is the assessment of suspected recurrence or distant metastases. A 2001 study by Kim et al showed 93% accuracy in identifying recurrent breast cancer (56). Another report demonstrated that PET performed better at detecting metastases than did bone scans (57). Because of the nature of osteocyte metabolism, PET detects osteoclastic lesions better than osteoblastic lesions. Bone scans favor osteoblastic processes because they detect the reaction of the bone to the malignant lesion. PET provides the additional benefit of imaging internal mammary nodes. Overall, Yap et al found that PET changed the mode of management in 28% of their patients (58).

PANCREATIC CANCER

Several studies have addressed whether PET is useful in identifying pancreatic cancer (59–63). In 434 patients, PET was 85% to 95% sensitive; however, specificity was 78% to 99% due to false-positive FDG uptake in some inflammatory lesions. Overall accuracy hovered between 85% and 93%.

If a surgeon is willing to perform a pancreaticoduodenectomy only in cases of potentially curable cancer, management may change in up to 41% of cases (27% malignant and 14% distant metastases) according to Rose et al (64). PET identified cancer in 18 of 60 patients who had indeterminate CT scans. Distant metastases were identified at the outset in 7 patients. Sperti et al also reported how PET influenced their management of cystic neoplasms of the pancreas (65). They were content to perform limited excisions of the cysts or simple biopsy alone in those patients who had negative scans.

With special techniques for quantification, parameters such as the hexokinase activity of the cell may be assessed by PET. Hexokinase may serve as an indication of proliferation, may be found to be a factor related to tumor grade, or may even be seen as a target for chemotherapy (66). Researchers from Germany recently tried to compare PET findings with those of CT, endo-scopic retrograde cholangiopancreatography, and intraoperative sonography (67). Inconsistency led the investigators to deem each of the techniques as unreliable, and they warned about the possibility of PET ushering in a new spectrum of “misinterpretation.” One particular limitation of PET is a false-positive scan in the setting of chronic pancreatitis (68). Again, optimal PET imaging of these tumors likely awaits approval of tracers for which activated white cells are less avid, such as nucleic acid or amino acid analogs.

LIVER CANCER

PET is not as well suited for evaluating hepatocellular carcinoma as it is for evaluating metastatic disease. Since cirrhotic livers have been found to have irregular glucose-6-phosphatase activity, the PET scan can be misinterpreted. In contrast, metastases are more easily found against a background of normal, relatively homogeneous hepatic parenchyma. Another source of error is the occasional intrahepatic abscess (69).

One potential role for PET is in the follow-up of tumor ablation. Chemoembolization efficacy has been considered (70). To date, however, there are no reports on using PET to monitor radiofrequency or ethanol ablation.

SARCOMA

In sarcomas, PET can be used to guide the surgeon toward areas of most intense mitotic activity and away from areas of necrosis. Such use may improve the accuracy of biopsy in assessing maximum tumor grade. Treatment progress may be followed by serial PET scans if there is doubt as to its efficacy. Misleading results may be obtained in cases of infection or hypermetabolic conditions such as Paget's disease.

NONTHYROID ENDOCRINE CANCERS

Carcinoid endocrine tumors have been shown to be poorly imaged by PET; consensus is that ¹¹¹In-Octreoscan is still the scan of choice in these conditions.

Incidentally discovered adrenal masses may accurately be classified as benign or malignant. According to a study by Boland et al, PET was 100% accurate, correctly classifying 14 malignant and 10 benign lesions (71). In patients who have a history of cancer, an adrenal mass represents metastasis in 27% to 36% of cases. PET is useful in such cases for identifying a distant metastasis, which may become a preferable target for a percutaneous biopsy. The anticipated benefit of radical surgery would have to be carefully evaluated.

SUMMARY

PET can be useful to the surgical oncologist in diagnosing tumors (particularly solitary pulmonary nodules), grading tumors (assessing baseline standardized uptake values and tumor metabolism), guiding biopsies of highly active tumor regions, staging tumors, monitoring treatment, and determining prognosis. PET may be an independent staging tool or be used for surveillance and in recurrence or restaging. PET has been clinically available for a relatively short time. We certainly will gain more knowledge regarding its appropriate role in the surgical management of oncologic patients as more experience is gained in a wide variety of tumor types.