Optimal timing of FDG PET during chemotherapy
Positron emission tomography (PET) with 18F-fluorodeoxyglucose (FDG) is successfully capable of imaging glucose metabolism of the tumor cells. Tumor glucose metabolism using FDG-PET has a potential to distinguish viable cancer cells from those in suspension or necrotic components because the degree of tumor FDG uptake is closely associated with its proliferation activity. There is cumulative evidence showing that reduction in FDG uptake value on the early phase after the initiation of chemotherapy more reliably predicts a favorable outcome of patients with breast cancer. Therefore, FDG-PET could help to individualize treatment and to avoid potentially ineffective chemotherapies. In this article, we discuss and illustrate the role and limitations of FDG-PET in the management of neoadjuvant chemotherapy in breast cancer.
2. Tumor metabolic response
FDG-PET has proven useful in the management of various cancers . FDG-PET is known to play an important role in the detection of distant metastasis and recurrence [2, 3]. In addition, it provides quantitative information on tumor glucose metabolic activity, allowing the measurement of metabolic changes and cancer activity shortly after initiation of therapy and before tumor volume reduction . This functional imaging tool may also be useful in predicting tumor response to therapy and optimizing individual treatment .
Tumor response to therapy is traditionally assessed by comparison of tumor size before and after treatment, which is determined using anatomical imaging devices such as ultrasonography and computed tomography (CT) on the basis of the Response Evaluation Criteria in Solid Tumors (RECIST) . Although RECIST has been widely adopted, tumor response must be evaluated for several months until surgery and it does not always reflect the pathological response because of difficulties in distinguishing residual cancer cells from necrotic lesions, fibrosis, or benign masses.
Because of limitations in applying the RECIST criteria using anatomic imaging alone, Wahl et al. proposed a draft framework called the PET Response Criteria in Solid Tumors (PERCIST) . The proposed PERCIST criteria are being used in several current clinical studies. As far as early response assessment with FDG-PET is concerned, clinical studies involving patients with Hodgkin lymphoma and aggressive non-Hodgkin lymphoma have demonstrated that the assessment of changes in FDG uptake on PET imaging after 2 or 3 cycles of chemotherapy was superior for predicting patient prognosis compared with the assessment of morphological changes on computed tomography (CT) . This method was shown to be at least as reliable as definitive response assessment at the end of therapy . In a European multicenter trial monitoring chemotherapy response and survival in 260 patients with lymphoma, an early decrease in FDG uptake on PET after 2 cycles of chemotherapy was significantly correlated with progression-free survival . For other types of cancers, including breast cancer , non-small cell lung cancer , esophageal cancer  , gastric cancer , and colorectal cancer , several clinical studies revealed evidence of the emerging role of FDG-PET in predicting both post-therapeutic clinicopathological response and patient survival.
In our institute, a discrepancy was observed between tumor morphological changes and tumor metabolic activity in a patient treated with neoadjuvant chemotherapy for primary breast cancer. FDG-PET was performed at the onset of chemotherapy, at the midpoint of chemotherapy, and prior to surgery. The primary tumor appeared to grow rapidly after the start of neoadjuvant chemotherapy despite the gradual reduction in glucose accumulation (Figure 1). Postoperative pathological analysis revealed that the lesion was replaced with scar tissue in addition to the presence of massive bleeding and small residual cancer cell nests. In this case, FDG-PET was able to provide more accurate and clinically beneficial information compared with CT.
Clinical studies conducted worldwide have repeatedly revealed the predictive value of FDG-PET in patients with advanced breast cancer treated by chemotherapy. As early as 1993, Wahl et al. studied 11 patients with locally advanced breast cancer before and after 1 cycle of chemotherapy. A significant difference was observed in tumor FDG influx rate (K) from baseline levels between responders and nonresponders (sensitivity: 100%, specificity: 100%) . In 1996, Bassa et al. conducted a retrospective study of 13 patients with breast cancer for whom FDG-PET scans were performed prior to chemotherapy, at the end of the first cycle, at the midpoint of chemotherapy, and before surgery. The mean standardized uptake value (SUV) of the tumor after the first cycle of chemotherapy was significantly lower than the baseline value (p < 0.01) . In 2000, reports of clinical studies from two separate institutes showed the usefulness of FDG-PET in the early evaluation of tumor metabolic response to chemotherapy. Schelling et al. demonstrated the ability of FDG-PET to differentiate between responders and nonresponders after the first course of chemotherapy (sensitivity: 100%, specificity: 85%) . Smith et al. also successfully utilized FDG-PET for predicting tumor response after the first cycle of chemotherapy (sensitivity: 90%, specificity: 74%) .
Our team investigated the maximum changes in SUV (SUVmax) in 32 primary breast cancer lesions in 30 patients. The patients were treated with neoadjuvant chemotherapy comprising 4 cycles of epirubicine and cyclophosphamide on a triweekly basis and sequential weekly cycles of taxane for 12 weeks . Figure 2 shows representative tumor images on FDG-PET performed at baseline, after one cycle of chemotherapy, after four cycles of chemotherapy, and prior to surgery. The serial images on the upper row show a tumor in which a pathological complete response (pCR) can be observed. The middle row depicts a tumor exhibiting a pathological partial response (pPR), and the lower row illustrates pathological progressive disease (pPD). SUV decreased dramatically in the pCR tumor after one cycle of chemotherapy, after which metabolic activity ceased. SUV change in the pPR tumor was lower than that in the pCR tumor after one cycle of chemotherapy, but SUV gradually diminished during further chemotherapeutic treatment. The pPD tumor showed no significant changes in SUV after treatment. In terms of the optimal threshold of a 40% decrease in SUV, the rate of pathological response in terms of pCR or near pCR was higher (71.4%) in metabolic responders than in nonresponders (12.5%). The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were 63%, 92%, 71%, and 88%, respectively.
In 2008, Schwarz–Dose et al. performed the first prospective multicenter trial to evaluate the effectiveness of FDG-PET in predicting early pathological response during chemotherapeutic treatment in 104 patients with locally advanced breast cancer . In that report, when a 40% decrease in SUV occurred in the first cycle after initiation of chemotherapy compared with baseline values, FDG-PET predicted pCR and pathological macroscopic residual disease at a high rate, with a sensitivity of 73%, specificity of 63%, PPV of 36%, and NPV of 90% . Other representative studies published in the literature since 2000 are listed in Table 1 [21-25].
In 2011, a meta-analysis in this field summarized 16 articles including a total of 920 patients with breast cancer . To predict histopathological response in primary lesions, the pooled sensitivity, specificity, PPV, NPV, and diagnostic odds ratios were calculated. The results for these parameters were 84% [95% confidence interval (CI), 78%–88%], 66% (95% CI, 62%–70%), 50% (95% CI, 44%–55%), 91% (95% CI, 87%–94%), and 11.90 (95% CI, 6.33%–22.36%), respectively. Although the checkpoints of FDG-PET administration in these trials differed, a subset analysis showed that early response of glucose metabolism after the first or second cycle of chemotherapy provided a significantly better indicator of accuracy compared with a later response (third cycle or later). The subset analysis results showed in Table1. These data indicate that changes in SUV during the first 1–3 cycles of chemotherapy are better indicators of clinical outcome compared with changes during the later cycles.
The high sensitivity and high NPV reported above indicate that FDG-PET may be useful for the identification of nonresponders among patients in the early phases of treatment with neoadjuvant chemotherapy. However, in cases where FDG-PET indicates changes in SUV (responders), decision-making regarding continuation of treatment may still be difficult. This problem has been addressed in research on lymphoma patients. Randomized trials have been conducted to determine whether response-guided treatment using early response to therapy as measured by FDG-PET scans is feasible or useful in decreasing the cumulative dose of potentially cytotoxic agents in nonresponders , . These trials aim at treatment modification based on PET response by comparing risk-adapted treatment guided by FDG-PET with standard chemotherapy in these patients .
An emerging paradigm for a treatment strategy using FDG-PET has been introduced in addition to traditional assessment of tumor response on the basis of staging and tumor subtyping . A specific treatment was chosen from a number of chemotherapeutic drugs, and the usefulness of FDG-PET was analyzed in comparison to assessment based on staging and subtyping. This new strategy to optimize treatment timing includes staging, subtyping, and response guiding. Staging and subtyping determine favorable treatment options before the initiation of treatment. Functional imaging with FDG-PET offers opportunities to assess tumor response early in the course of treatment. This response-guiding strategy offers the opportunity to revise treatment and improve outcome. FDG-PET not only provides invaluable prognostic information in patients  but also supports efforts to switch to more effective treatment options in the early stages of treatment rather than on completion of therapy (Figure 3).
3. Limitations of FDG-PET
Although tumor FDG uptake is an indicator of the viability of cancer cells , , it is influenced by many biological factors such as stromal cell activity , tumor perfusion , immune reaction , hypoxia , , and apoptosis. The acute effect of cytotoxic drugs on tumor FDG uptake occurs as a result of high glucose uptake by inflammatory cells and/or energy demand in the process of acute apoptotic death , leading to a transient increase in tumor FDG uptake (so-called “flare response”) , . Therefore, some authors have claimed that the timing of PET scanning soon after the onset of chemotherapy treatment may be crucial. To avoid the flare response, scanning must be delayed for at least 1–2 weeks after the initiation of chemotherapy . PET may be administered immediately before initiation of the second cycle of chemotherapy.
One cause of confusion regarding the results of FDG-PET is the heterogenous results concerning the pathological criteria of outcome, which distinguishes responders from nonresponders. The overall prognosis of patients remains unconfirmed. In addition, variations in dose and different combinations of drugs add to the confusion. The optimal timing of PET scanning may depend on dose intensity or regimen. If the regimen changes during the course of treatment, the results may be affected because the mechanism of drug sensitivity differs according to tumor characteristics.
The optimal timing of FDG-PET scanning after initiation of chemotherapy for the prediction and elimination of progressive disease (PD) has also not been determined till date. In clinical practice, most physicians are more concerned about tumor progression during neoadjuvant chemotherapy than about achieving pCR. Because chemotherapeutic treatment may increase genomic instability in some tumors, and because chemotherapy resistance may develop in severe cases with hypoxia, neoadjuvant chemotherapy may be contraindicated. However, the role of FDG-PET in the early prediction of PD remains to be established.
The optimal threshold of metabolic change on FDG-PET also remains unclear. A recent draft standard of PERCIST has been advocated as response criteria in solid tumors . It recommends a ≥30% decrease in SUV as a cutoff value for partial metabolic response, which is associated with clinical outcome after chemotherapy. However, specific response criteria for breast cancer must be defined in order to increase the accuracy of prognosis. Further prospective research is needed to determine the optimal cut-off value for predicting tumor response.
Thorough evaluation of cost-effectiveness and prognostic impact of early switching from ineffective neoadjuvant chemotherapy to a more effective regimen is essential. In a simulation study, Schegerin et al. found that early prediction of tumor response using functional imaging devices such as FDG-PET facilitated tailoring of treatment options, which had economic benefits . Further clinical trials must be conducted in order to shed further light on this topic.
Finally, the European Organization for Research and Treatment of Cancer PET study group recommended improvements in the quality of tumor imaging with FDG-PET. Interinstitutional bias is another factor in the usefulness of FDG-PET as a prognostic tool. The available data are insufficient to define the optimal time after injection of FDG and the optimal dose of FDG at which SUV should be measured .
4. Early changes in metabolism using molecular-targeted drugs and endocrine therapy
Functional imaging devices may be useful in the assessment of the biological activity of molecular-targeted drugs . These agents are predominantly cytostatic in nature, that is, they modulate biological behavior and arrest cell cycling rather than totally killing cancer cells . In cases treated with these agents, the traditional endpoints used to evaluate the effects of cytotoxic drugs, such as RECIST criteria, are insufficient and sometimes inappropriate for the prediction of therapeutic outcome. For example, antagonists of the epidermal growth factor receptor (EGFR), such as trastuzumab and cetuximab, block membranous EGFR in cancer cells, halt cell cycling, and induce apoptosis. These drugs and many others require different evaluation criteria to determine their efficacy. FDG-PET may be useful in establishing these criteria.
Small molecule drugs to block protein kinases, such as gefitinib, have been used for the inhibition of tumor proliferation and neovascularization. A rapid decrease in FDG uptake at 48 hour was seen in lung cancer xenografts treated with gefitinib. Su et al. also reported a very early decrease at 2 hour in gefitinib-sensitive cancer cells and no change in resistant cancer cells . Other reports stated that a reduction in metabolism within 1 week after the commencement of therapy was associated with sensitivity to certain drugs, for example, an epidermal growth factor receptor/human epidermal growth factor receptor 2 (HER2) dual kinase inhibitor (lapatinib) used for the treatment of breast cancer, a c-kit inhibitor (imatinib mesylate) used for the treatment of gastrointestinal stromal tumors (GIST), a mammalian target of rapamycin inhibitor (rapamycin) used for the treatment of GIST, and various other drugs used for the treatment of uterine and neuroendocrine carcinomas and sarcomas . Moreover, in the clinical setting, FDG-PET was reported to be useful for the evaluation of treatment response to sunitinib, a multitarget tyrosine kinase inhibitor, in patients with GIST resistant to treatment with imatinib .
Endocrine therapy is one of the most common treatment strategies in patients with estrogen receptor (ER)-positive breast cancer. Successful cytostatic drugs include ER antagonists such as tamoxifen, which induce the deprivation of estrogen production, aromatase inhibitors, ER downregulators, or fulvestrant. In 2011, we reported that changes in SUV at approximately 2 weeks after treatment with letrozole, an aromatase inhibitor, was correlated with a drop in proliferative rate of cancer cells measured by immunohistochemical staining of Ki67 . With a tentative threshold value of a 40% decrease in SUV, Ki67 index values were significantly decreased in metabolic responders. The Immediate Preoperative Anastrozole, Tamoxifen, or Combined with Tamoxifen (IMPACT) randomized trial revealed that 2 weeks of treatment with the aromatase inhibitor anastrozole suppressed the Ki67 index (as compared with a percentage of baseline expression) to a significantly greater extent than did tamoxifen alone or tamoxifen in combination with anastrozole. The affiliated study showed that after 2 weeks of endocrine therapy, Ki67 index values predicted recurrence-free survival in individual patients . A positive correlation between the Ki67 index and tumor SUV has been reported in some studies [32, 49], . Mortazavi-Jehanno et al. investigated the predictive value of metabolic response in patients with metastatic breast cancer after 8 weeks of endocrine therapy, demonstrating that progression-free survival is related to metabolic response . These observations suggest that changes in tumor SUV after endocrine therapy may be associated with favorable prognosis. Therefore, the biological basis of changes in FDG uptake using cytostatic drugs may be associated more with intracellular pathways of metabolism and cell cycling than with cytotoxic agents.
4.1. Tumor metabolic flare
Tumor flare reaction denotes a sudden and temporary worsening of tumor-related symptoms after the initiation of treatment . Several studies reported that radiotherapy and some types of chemotherapeutic agents induce diffusely elevated FDG accumulation because of inflammation. Weber suggested a careful inspection of the degree and pattern of FDG uptake to distinguish between radiation-induced inflammation and residual cancer activity . As mentioned earlier, the EORTC PET study group recommends that after baseline FDG-PET scanning and before the initiation of chemotherapy, serial scanning using FDG-PET should be performed 1–2 weeks after the first course . Therefore, the consensus till date has been that a waiting period of 1–2 weeks should be observed after initial drug administration or radiotherapy in order to avoid the inflammatory response and accurately evaluate tumor activity. However, metabolic flare does not necessarily indicate treatment failure or cancer progression. Table 2 lists the imaging studies reporting the association between tumor metabolic flare and tumor response to treatment.
In animal experiments, Furuta et al. reported a flare reaction detected by FDG-PET in nude mice with ependymoblastoma, small cell lung cancer, and glioblastoma at 2 h after irradiation with 10 Gy . They claimed that flare intensity was strongest in the ependymoblastoma, the most radiosensitive of these tumors, whereas the two less radiosensitive tumors showed no increase in FDG uptake during the observation period. In human testicular cancer xenografts in nude rats that received cisplatin, Aide et al. reported that FDG-PET scanning detected a peak FDG uptake on day 2, followed by a marked decrease on day 7 despite the lack of change in tumor volume . They observed a transient S and G2/M cell cycle arrest and a marked increase in apoptosis within this phase. A very early increase in FDG uptake may explain the flare reaction that represents increased tumor metabolism in apoptotic cells as well as in cells that exhibit transient cell cycle arrest.
Aliga et al. reported a similar result mice after the administration of doxorubicin to decrease tumor burden in BALB/c mice with breast cancer . They observed a rapid decrease in tumor FDG uptake 24 hour after chemotherapy and a transient accumulation in FDG uptake on day 7. They suggested that a partial agonistic effect of chemotherapy, apoptosis, cell repair mechanisms, or intratumoral inflammation may be responsible for this flare reaction.
Using the fluorescent glucose analog 2-NBDG, Bjurberg et al. reported a very early increase in metabolism in three squamous cell carcinoma cell lines (LU-HNxSCC-7, LU-HNxSCC-24, and LU-CX-2) after exposure to cisplatin  . The flare reaction was observed within 3 days after exposure in these cells, whereas FDG uptake in nonmalignant fibroblastic cells was low.
Contrary to recent animal studies revealing a flare phenomenon-associated tumor response, only a few clinical studies have utilized FDG-PET in detecting tumor flare. As early as 1996, Schneiders et al. reported the use of scintigraphy to detect a flare reaction in bone lesions in patients with metastatic disease; the flare reaction occurred despite a favorable overall outcome . This phenomenon may represent enhanced osteoblastic activity, rapid bone repair, and improved blood flow around the responding lesions. Welch et al. conducted an FDG-PET study in which a paradoxical flare phenomenon was detected within 7 to 10 days after the initiation of tamoxifen in patients with breast cancer . An increase in FDG uptake was observed in responding tumors during week 1 after tamoxifen initiation ; however, this was not observed in nonresponding tumors. The transient increase in FDG uptake could reflect hormone receptor-related changes in tumor metabolism, which was predictive of a favorable outcome. Partial agonist–antagonist activity in selective ER modulators such as tamoxifen is known to differ over the treatment course after tissue exposure. The slow onset of action of tamoxifen and the fact that its effects as an estrogen agonist peak 1–2 weeks after the onset of therapy may contribute to the development of the flare reaction. The same group of investigators conducted a clinical study including an estradiol challenge test to predict hormone sensitivity in women with locally advanced or metastatic ER-positive breast cancer. FDG-PET after 30 mg estradiol induced a metabolic flare, showing greater responsiveness to endocrine therapy and better overall survival in flare patients than in non-flare patients .
Results of those animal studies lead to the speculation that tumor metabolic flare is related to the following sequence of events: transient cell cycle arrest, apoptosis, induction of the cancer immune system, and hemodynamic reaction. As for estrogen-positive breast cancer and endocrine therapy, estrogen stimulation may be closely associated with an increase in glucose uptake . At lease, this does not necessarily indicate refractory disease.
Taken together, the underlying presumption in these studies is that changes in FDG uptake in response to chemotherapy occur in three phases (Figure 4). In the first phase, cellular damage followed by inflammation and vascular changes occurs within hours or days, resulting in accumulation of FDG and increased uptake in cancer cells and inflammatory cells such as neutrophils (inflammatory phase). In this phase, pro-inflammatory cytokines may be released because of the tumor’s response to chemotherapy, and these cytokines accelerate tumor cell proliferation, activate the immune system, and increase blood flow, resulting in accumulation of tumor FDG. Second, apoptosis, cell cycle arrest, and diminishing inflammation, which occur within days or weeks, lead to a decrease in FDG uptake in cancer cells (apoptotic phase). Decreased blood flow due to vascular damage caused by chemotherapy may deteriorate the tumor microenvironment and promote apoptosis. In addition, tumor cells may be replaced by fibrotic cells, shrinking the tumor several weeks after the initiation of chemotherapy. FDG uptake in the tumor then decreases because of volume reduction (volume reduction phase). In this last phase, morphological changes can be identified by CT or magnetic resonance imaging. Therefore, metabolic flare precedes decreased uptake of FDG. This is not necessarily a confounding factor; rather, it provides an insight into pharmacodynamics . Further understanding of cancer metabolic flare in the early phase after chemotherapy can aid in strategic planning of successful therapy.
Of late, cancer treatment is frequently optimized on the basis of tumor subtype and stage. Breast cancer is characterized by ER and HER2 status in addition to tumor size and distant metastatic involvement. The advent of commercialized kits containing multiple molecular biomarker assays has shifted the focus in tumor categorization from pathology to molecular analysis. However, rapid development and wide availability of a broad range of new drugs has exceeded the discovery of tumor subtyping methods or the establishment of biomarkers predictive of chemosensitivity. Therefore, a new paradigm that involves response-guided strategies during initial treatment, which can aid in decision-making about the next treatment option, is emerging for cancer treatment. Early response assessment using FDG-PET may eventually be applicable for planning and evaluating future treatment strategies. In future, patients could be allocated to standard or investigational chemotherapy regimens on the basis of metabolic response. Furthermore, in phase 1/2 studies determining the optimal dose of a new drug, nonresponsive patients can be eliminated on the basis of metabolic response. Glucose metabolism analysis using FDG-PET will be one of several critical factors for evaluating tumor response to chemotherapy. Integration of multiple functional imaging systems may also be useful in predicting early tumor response to chemotherapy.
Finally, considering the phenomenon of tumor metabolic flare, clinical trials must be conducted to determine the best timing for the administration of FDG-PET. This information would be useful to predict tumor response in the very early stages of treatment.
We dedicate this review article to the late Dr. Yoshiaki ABE for helpful cooperation with a series of PET studies. We are very grateful to Dr. Hitoshi TSUDA of the National Cancer Institute for his helpful advice and to Dr. Jiro ISHIDA and Dr. Katsumi TAMURA from the Tokorozawa PET Diagnostic Imaging Center for the clinical reading of FDG-PET images.