Applied Radiology

Positron emission tomography (PET) is a noninvasive nuclear medicine study that has been in existence for almost 30 years, but has been gaining acceptance in oncologic imaging rapidly during the past 5 years. It involves injection of a shortlived positron-emitting radiopharmaceutical, 2-deoxy-[F-18]fluoro-Dglucose (FDG), a glucose analog with an approximate 2-hour half-life. When F-18 decays, it emits a positively charged electron, or positron, which travels a few millimeters in soft tissue before combining with an electron. Two high-energy "coincident" gamma photons are emitted at 180 degrees apart from each other as a result of this positron-electron annihilation reaction and are detected by a circular array of detectors in the PET scanner as they exit the patient's body.

The rationale for use of this radiopharmaceutical is that most malignant lesions have accentuated glucose metabolism, which is mirrored by increased uptake of FDG. Since FDG cannot be metabolized within the cell like glucose, it is effectively "trapped" within cancer cells. Malignant lesions appear visually as areas of increased activity ("hot spots") on a PET scan. PET images are analyzed routinely by qualitative visual methods, but also are often analyzed semiquantitatively, using the "standardized uptake value" (SUV). This method assigns a numerical value to the intensity of FDG uptake within a region of interest incorporating the neoplasm. The SUV relates the activity concentration in a certain volume of tissue to the amount of the injected dose and the patient's body weight. In general, malignant lesions have an SUV in the range of 2.5 to 15.

Other radiologic studies, such as mammography, sonography, CT, and MRI, provide detailed anatomic information about the size and location of masses, but not the unique metabolic information available with PET. This metabolic information generally affords PET several advantages over the anatomic modalities, including: earlier detection of malignancy; differentiation of scar or benign lesion from active malignancy; detection of metastatic disease in normal-size lymph nodes; and assessment of early tumor treatment response.

Breast cancer is the most common non-dermatologic malignancy in women in the United States, where approximately 192,000 women were diagnosed with breast cancer in 2001, with 40,000 mortalities. The incidence is increasing and a woman now has a 1 in 8 chance of developing breast cancer in her lifetime.1 Accurate staging of breast cancer has important therapeutic and prognostic implications for optimal patient care.

Recently, the Centers for Medicare and Medicaid Services (CMS), previously known as HCFA, considered the utility of PET in breast cancer, after granting approval for the use of PET in non-small-cell lung cancer, colorectal carcinoma, lymphoma, esophageal cancer, melanoma, and head-and-neck-cancer since 1998. They commissioned the Blue Cross and Blue Shield Association (BC/BS) Technology Assessment Center, an evidence-based practice center, to evaluate the pertinent literature on PET in breast cancer and also polled PET experts and clinical oncologists with PET experience. Four clinical applications of PET in the evaluation of breast cancer were considered by CMS: 1) initial diagnosis of breast cancer; 2) initial staging of axillary lymph nodes; 3) detecting locoregional recurrence or distant metastases; and 4) evaluating response to treatment.

INITIAL DIAGNOSIS

Mammography is widely accepted as a screening test for breast cancer, partly because it can detect cancer in nonpalpable lesions, is inexpensive, and is widely available. The accuracy of mammography is much lower in women on hormonal replacement therapy, as well as those with dense breasts, implants, fibrocystic disease, and prior breast surgery. The sensitivity of mammography for detection of cancer was only 30% in women with very dense breasts and 60% in those with heterogeneously dense breasts.2 In a large population-based database of approximately 183,000 patients, screening mammography had an overall sensitivity of about 80%, again, this figure was lower in younger women with denser breasts.3 Low specificity is another limitation of mammography. More than 80% of suspicious microcalcifications are histologically benign, necessitating a large number of unnecessary biopsies. Due to the limitations of mammography, additional imaging with sonography and color Doppler sonography have been used as adjunctive tests in breast imaging. While they improve the specificity of cancer detection, they cannot detect microcalcifications, which are often the only sign of malignancy. MRI also has been used in difficult mammographic cases, especially in dense breasts. Although it has better sensitivity in these cases, it still has a low specificity of cancer detection.4

The decision memorandum by CMS on use of PET scanning in the initial diagnosis of breast cancer5 notes that 13 studies, for a total of 606 subjects, met their selection criteria. A meta-analysis of these studies found a sensitivity of 89% and specificity of 80% for PET in the diagnosis of breast cancer. They noted flaws in most of these studies, however. The most significant flaws cited were small study sizes, large mean tumor sizes (2 to 4 cm), and the high prior probability of malignancy in subjects studied with PET. There is, indeed, insufficient literature on PET's sensitivity in detecting breast lesions smaller than 1.5 cm with current state-of-art PET technology (Figure 1). CMS chose not to approve PET for detection of breast cancer, citing uncertain applicability of PET data to populations with smaller tumor sizes and a lower probability of malignancy. Nevertheless, it is highly likely that PET will play a significant role in the evaluation of such patients in the future. This field is likely to be advanced by development of dedicated PET instrumentation for imaging the breast and axilla.

INITIAL STAGING OF AXILLA

 Staging of axillary lymph nodes has been established as an important prognostic indicator in breast cancer. Anatomic studies have been inadequate in evaluating the axilla, primarily due to the presence of metastatic disease in normal-size lymph nodes. Extensive surgical sampling of axillary nodes, with its attendant morbidity, has been utilized in the past for accurate staging. More recently, limited axillary dissection has become feasible, using the lymphoscintigraphic and/or blue dye "sentinel node" method. A variety of studies have assessed FDGPET for initial staging of axillary lymph nodes. In general, those with a high sensitivity (>90%) have had a lower specificity (66% to 89%).6-8 Those studies in which the specificity was maximized to >90% had a lower sensitivity (79% to 90%).9-12 Smith et al,12 stated that PET is the most accurate noninvasive method for assessing the axilla in breast cancer (Figure 2). However, it is probably not sufficient to replace surgical sampling of axillary nodes because of the important prognostic implications of accurate staging. In order to test PET's effectiveness for occult disease, the BC/BS technology assessment searched for papers that included patients with confirmed breast cancer, no palpable axillary lymph node metastases, and no evidence of distant metastases. Only four studies satisfied these criteria. In this small aggregate group of 203 subjects, a meta-analysis yielded a pooled sensitivity of 81% (range 40% to 93%) confirming that the detection of micrometastases and small tumor-infiltrated lymph nodes is limited by current PET resolution. The specificity of the pooled data was higher (95%, range 87% to 100%). The BC/BS committee felt the available data were too sparse to draw an appropriate conclusion and that the false-negative rate (19%) was too high, which would result in undertreatment of too many patients with local metastatic disease.5 CMS chose not to approve PET for axillary staging per se, meaning that axillary node sampling should remain the standard of care. Nevertheless, current data clearly indicate that, when PET is performed for overall staging of metastatic disease in patients with breast cancer, it can provide clinically valuable information regarding axillary nodal stage in many cases.

DETECTION OF LOCOREGIONAL RECURRENCE OR DISTANT METASTASIS/RECURRENCE (STAGING AND RESTAGING)

When the diagnosis of breast cancer is made, accurate staging is important for optimizing treatment and estimating prognosis. A battery of diagnostic tests often includes CT or MRI of the brain, bone scintigraphy, chest radiography and sonography, or CT of the abdomen. One of the established advantages of PET over anatomic imaging for other cancer types is the ability to characterize lesions discovered incidentally on anatomic studies as malignant versus benign. Another is its ability to evaluate a large portion of the body with one test (Figure 3). There are few reports in the literature comparing PET with conventional imaging methods for breast cancer staging or restaging. In one report, the accuracy of PET was compared with CT and MRI in 75 patients with suspected recurrent or metastatic disease.13 PET performed well in this small series, detecting 28 of 29 patients with lymph node disease, 15 of 15 patients with bone metastases, 5 of 6 patients with lung metastases, and 2 of 2 patients with liver metastases. PET detected 8 lymph node metastases and 7 bone metastases that were not detected by CT or MRI. Hubner et al14 found PET to have an 85% sensitivity and 73% specificity versus 71% and 54%, respectively, for CT. In another series of 57 patients, all referred because of a suspicion of disease recurrence, sensitivity and specificity of PET on a per patient basis were 93% and 79%, respectively.15 On a per lesion basis, the sensitivity and specificity values were 85% and 79%, respectively. A disproportionate number of false-negative exams involved osseous metastases. When only nonosseous metastases were included in the lesion analysis, the sensitivity rose to 96%. Lower sensitivity for detection of bone metastases has been noted in other series. Cook et al16 found that bone scintigraphy is more accurate than PET for osteoblastic bone metastases, while PET is more accurate for osteolytic metastases. It is also well established that PET is not sufficiently sensitive for detecting brain metastases. This is attributed to the high background FDG activity in cerebral gray matter, obscuring small metastases.17 Ultimately, PET may be able to replace some of the battery of currently used anatomic staging tests, but it will likely not replace MRI of the brain and whole body bone scintigraphy for these reasons.

Detection of local recurrence is especially difficult after surgery and in patients who have had breast augmentation mammoplasty because of the anatomic changes related to scarring. PET is useful for differentiating scar from tumor in many types of cancer, including breast cancer (Figure 4). Bender et al13 correctly identified 16 of 20 patients with local recurrence with PET. An additional potential use for PET is in detecting occult mediastinal and internal mammary nodal metastases. This can be difficult with CT, due to its reliance on nodal size criteria. CT cannot detect metastatic disease in normal-size nodes and cannot differentiate enlarged, reactive nodes from metastatic nodes. In a recent study of 33 patients suspected of having only locoregional disease, 10 had unsuspected mediastinal or internal mammary disease detected by PET.18

 A recent paper approached the comparison between PET and conventional imaging (CI) in the assessment of suspected recurrent breast cancer from the point of view of prognosis.19 The sensitivity and specificity of PET for recurrent breast cancer were 93% and 84%, respectively, while those for CI (a combination of radiography, bone scintigraphy, CT, MRI, mammography, and sonography, selected for each patient based on routine clinical management parameters) were 79% and 68%. The negative predictive value of PET was 80% for PET versus 59% for CI. PET also performed significantly better than CI in predicting disease-free survival in these patients.

 CMS cited limitations in the PET literature for detecting recurrent locoregional disease and distant staging. Nevertheless, they felt that the evidence was sufficient to conclude that "PET could have a positive adjunctive role when used with standard imaging technology" and approved coverage of PET for this indication.5

EVALUATING TREATMENT RESPONSE

Another potentially important role for PET scanning is in monitoring early treatment response. Theoretically, earlier recognition of ineffective therapy could allow a change to an alternative, and hopefully more effective, chemotherapy regimen. Treatment monitoring with PET can include either neoadjuvant chemotherapy for locally advanced primary breast cancer or treatment of distant metastatic disease. Although there is even less supporting literature for these indications than for other applications of PET in breast cancer, there are no other good alternatives for monitoring early treatment response.

 The goal of neoadjuvant chemotherapy is to reduce the size of the primary neoplasm in order to enhance the likelihood of successful primary resection and, possibly, permit a breastconserving surgical approach. Current standard imaging tests, such as mammography, sonography, CT, and MRI are hampered by a prolonged lag time of weeks to several months before anatomic changes are measurable. Furthermore, even when anatomic changes occur, scar cannot be differentiated from viable tumor, often necessitating histopathologic sampling. Studies have shown that metabolic imaging with FDG-PET is more effective than anatomic imaging in monitoring early treatment response. A rapid decrease in glucose metabolism in responders can be detected on PET as early as after the first cycle of chemotherapy.20-22 These studies utilized serial measurements of SUV. Successful local treatment can be documented by a decreasing SUV on serial PET scans, as much as a 55% decrement after the first course of chemotherapy in one study.21

Monitoring therapy of distant metastatic disease is even more difficult, since tissue sampling may not be feasible without significant morbidity. Earlier assessment of treatment response could be beneficial to guide further therapy and prevent prolonged treatment with ineffective drugs that have potent side effects. Until now, anatomic imaging with CT or MRI has been the standard-ofcare to assess treatment effect of distant metastatic disease, though data regarding the efficacy and, especially, the prognostic value of these techniques are extremely limited. As with neoadjuvant therapy, this approach has been suboptimal because anatomic changes are slow to reflect treatment response. Metabolic changes, however, occur much earlier and can be assessed with PET (Figure 5).

Despite the relative scarcity of supporting literature for monitoring treatment of breast cancer with PET, CMS was influenced by the drawbacks of conventional imaging for this purpose. Consequently, they approved coverage of PET for this indication. We await final guidelines (which may vary by regional and local Medicare carriers) regarding the approved timing of such follow-up studies. In the case of the tumors for which CMS has previously approved PET for follow- up (re-staging), such studies are only allowed after the completion of a course of therapy, when PET is likely to affect further treatment decisions. Whether CMS will allow interim PET follow-up during a course of therapy for breast cancer is not known at the time of this writing.

FINANCIAL CONSIDERATIONS

FDG-PET is an expensive technology. Current PET scanners can cost from $1 to $1.75 million. Medicare currently reimburses approximately $2000 for a wholebody PET scan, and reimbursement for hospital outpatient imaging will decrease somewhat this year. This reimbursement includes the cost of FDG, which varies regionally across United States from $250 to $800, depending upon the proximity of the PET center to a cyclotron facility. Nevertheless, PET can be cost-effective. Since it can differentiate scar tissue from tumor and distinguish many benign lesions from malignant ones, a negative PET scan can, theoretically, offer a very cost-effective alternative to surgery/ biopsy. This has already been demonstrated for lung cancer, colon cancer, and melanoma,23 in which prevention of surgery saved thousands of dollars per patient. Preventing unnecessary morbidity cannot be measured in dollars.

A positive PET scan is often confirmed histologically. PET has the potential to find the most easily accessible site for needle biopsy confirmation, again, reducing morbidity and the cost of an open biopsy.

 Monitoring chemotherapy with PET also can be cost-effective. Despite advances in treatment, some therapeutic regimens are still very toxic and expensive. By providing more accurate and much earlier assessment of treatment response, PET could, potentially, reduce the cost and morbidity of ineffective or unnecessary drugs.

  CONCLUSION

Although its current expense precludes PET from becoming a screening test for breast cancer, technological improvements will ultimately improve PET's spatial resolution, potentially allowing for the reliable detection of subcentimeter cancers. It may then become a valuable complement to mammography and breast ultrasound in problematic cases. Similarly, PET may never be adequate to replace initial axillary lymph node sampling, because it may never be able to reliably visualize microscopic metastatic disease that can be detected with thorough histopathologic evaluation. However, it should be remembered that PET has been shown to be the most accurate noninvasive imaging modality for assessment of axillary nodes, meaning that important information on axillary staging can be obtained with PET in certain patients, especially those undergoing follow-up after limited surgery or therapy.

There are no optimal noninvasive alternatives to PET for detecting recurrent breast cancer in the chest wall, for early treatment monitoring of either neoadjuvant chemotherapy, or for monitoring treatment of distant metastases. PET also has been demonstrated consistently to be more accurate than alternative imaging tests for staging distant disease in breast cancer, excluding brain and osteoblastic bone metastases. Assuming that it will be accepted in a similar fashion to the other approved oncologic indications, PET will soon be an important new imaging option for oncologists treating patients with breast cancer.