Positron Emission Tomography in Renal Cell Carcinoma: An Imaging Biomarker in Development

Abstract

Positron emission tomography (PET) has revolutionized cancer imaging. The current workhorse of molecular imaging, fluorodeoxyglucose (FDG) PET is used in the majority of malignant tumors with a few exceptions. Renal cell carcinoma (RCC) is one of those exceptions because of its variable uptake of FDG, although this variable uptake may actually be an asset in predicting response to some targeted agents, as will be discussed later. Beyond FDG, there is only scattered information in the literature on the use of PET in RCC. The purpose of this review is to summarize the current status of PET usage in RCC and point out its potentials and future directions.

We will start with a brief overview of the demographics, molecular pathogenesis, and evolving treatment strategies in RCC because this information is essential for better understanding of uptake of various PET radiotracers in this cancer and their indications. This will be followed by discussing the role of PET in characterization of indeterminate renal masses, in staging and restaging of RCC, and, finally, in predicting and monitoring therapy response. Each of these 3 areas of PET usage will include the relevant radiotracers currently in use or in development.

Demographics, Molecular Pathogenesis, and New Treatment Strategies in RCC

Cancers of the kidney make up an increasing portion of the cancer demographic. Annually, about 58,000 patients are diagnosed with cancers of the kidney, and roughly 12,000 die as a result of kidney cancer.¹ Primary malignancies of the kidney encompass a wide range of histologically different cancers, the majority (approximately 95%) being categorized as RCC.² Other tumors originating in the kidney include urothelial cancers, sarcomas, small-cell carcinomas or other neuroendocrine tumors, and lymphomas; but collectively, these tumor types make up <5% of the cancers of the kidney.³ Within the group of RCC, which is the focus of this review, about 80% of the tumors are of clear cell histology (ccRCC), and the remainder are made up of papillary or chromophobe variants, or some mixture thereof. RCC patients typically present in their seventh decade of life, and therefore, with the steady rise in incidence of this cancer (3% per year for the past 30 years in United States) and the aging nature of our population, make up an increasing proportion of the cancer patients in the United States and elsewhere.⁴

The characteristic molecular finding of ccRCC, the most common type of RCC, is the absence or nonfunctionality of the tumor suppressor protein, von Hippel–Lindau (pVHL).^5,6 pVHL functions as part of an E3 ubiquitin ligase that ubiquitylates a family of proteins known as the hypoxia-inducible factors (HIF), targeting them for degradation by the proteasome. In the absence of functional pVHL, HIF proteins, including HIF-1α and HIF-2α, which act as transcription factors, are allowed to accumulate. Transcriptional targets of HIF include a large number of tumor-associated genes, such as vascular endothelial growth factor (VEGF), glucose transporters (Glut), hexokinase, and carbonic anhydrase IX (CAIX). The fact that the effect of some targeted drugs used in RCC (eg, VEGF receptor inhibitors) and the uptake of some PET radiotracers (eg, FDG, ¹²⁴I-cG250) are dependent on HIF signaling makes PET an interesting biomarker for in vivo evaluation of this malignancy. At the same time, it should be noted that expression of markers that are important in PET imaging, such as Glut or CAIX, are increased in ccRCC because of lack of pVHL function and the resultant “pseudohypoxic” state, and not because of true hypoxia. This detail is important for understanding of mechanisms behind uptake of various PET radiotracers, or lack thereof, in ccRCC.

ccRCC presents in about 30% of patients as primary metastatic disease, and another 30% develop metastases after excision of primary localized disease, needing systemic therapy. Metastatic or recurrent ccRCC carries a grave prognosis, with a 5-year survival rate of <20% despite recent advancements in treatment and diagnosis.^7,8 The treatment paradigm for metastatic RCC has shifted in the recent years away from immune-based therapies in favor of molecularly targeted agents, primarily those targeting VEGF signaling or the mammalian target of rapamycin (mTOR) complex 1. Taking into account that the management of RCC has become more complex because of numerous therapeutic options, PET has potential to be extremely useful if it could a priori determine which RCC patients would likely benefit most from a particular targeted agent, and to monitor the effectiveness of the agent, once treatment has started.

Two important decision points in treatment of patients with renal masses are (1) determining the histology and malignant potential of a renal mass before planning surgery or other management and (2) determining the metabolic profile of primary and metastatic lesions when they present to predict behavior and outcome in response to metabolically targeted therapy. The first decision point is critical because surgical management for many renal masses, even when they are malignant, may be unnecessary or associated with excessive morbidity. The second decision point is growing in importance because there are efforts underway to match targeted drugs to specific tumor metabolic profiles. There are reasons to believe that PET could play an important role in both of these decision points to complement or replace tissue-based examination with noninvasive imaging-based examination. This is even more important not only because of the inevitable invasive nature of procedures that associated with tissue sampling but also because of the biases introduced into diagnostics confounded by sampling error and also due to the altered biology of the tissue by the time it arrives under the microscope. These changes occur because of type and duration of anesthesia, presence and duration of tissue ischemia after renal artery clamping, and tissue storage conditions (ie, medium and temperature), to name only a few.⁹ Because the development of new targeted drugs for RCC (and other malignancies) was not paralleled by appropriate evaluation methods for these kinds of drugs, new predictive and monitoring biomarkers are urgently needed.¹⁰ Currently, computed tomography (CT) and RECIST are most commonly used to evaluate treatment effects of tyrosine kinase inhibitors and other targeted drugs. However, size changes associated with these drugs are modest at most. Efforts to incorporate densitometry measurements have similarly been met with difficulty in consistent evaluation.¹¹ Although development of new PET radiotracers in form of labeled antibodies (eg, ¹²⁴I-cG250) and small molecules (¹²⁴I bevacizumab) advances rapidly, the potential of some of the currently available radiotracers, such as ¹⁸F-FDG and ¹⁸F-fluorothymidine (FLT), to predict and monitor the response to targeted drugs has not been completely exploited. We hope that this review will help to develop PET as an imaging biomarker in RCC.

PET in Characterization of Indeterminate Renal Masses

Nowadays, renal masses are mostly detected as incidental findings on cross-sectional imaging studies performed for other purposes. Not only because of the uncertainty and anxiety that such a finding can cause, an incidentally discovered renal mass cannot be considered a simple renal cyst, requiring prompt characterization, if any of these features are present: calcification, septations, multiple locules, wall thickening, nodularity, high attenuation (>20 HU) on unenhanced CT, signal intensity not typical of water at magnetic resonance imaging (MRI), or enhancement on contrasted studies.¹²

Renal protocol CT is the modality of choice in characterizing renal masses with a sensitivity of >90% even in smaller masses.¹³ The current standard multidetector CT protocol for evaluation of a renal mass consists of noncontrast and postcontrast acquisitions, the so-called “renal protocol CT.” Lesion size, shape, and borders can be evaluated on postcontrast CT, but enhancement is the most important factor in determining the likelihood of malignancy. CT enhancement positive for renal malignancy is defined as an attenuation increase of at least 15–20 HU from the corresponding noncontrast image. Lesions demonstrating an increase of <10 HU are generally considered as more likely to be benign, whereas lesions with an enhancement of 10–20 HU are not definitively categorized as benign because these may represent less enhancing variants of RCC, such as papillary type, and need to be further characterized based on their other features on CT or MRI. Despite high sensitivity, contrasted CT has a limited specificity for detecting surgical candidates because lymphomas and metastases from other primaries, indolent tumors with limited metastatic potential, such as papillary and chromophobe variants of RCC as well as benign tumors, such as oncocytoma and low-fat angiomyolipoma, can display similar enhancement features.

¹⁸F-FDG-PET

There are many publications evaluating the role of FDG-PET in characterization of renal masses. However, a closer look indicates that most of these studies did not really examine “truly indeterminate” renal masses and instead included patients who were already presurgical, that is, mostly based on renal-protocol CT, there had been reasonable concern that these masses could be malignant, before including them in the study. Furthermore, many of them were retrospective, and cases were selected from the PET archive and not from the population of the patients presenting with indeterminate renal masses. Therefore, data from the majority of these publications cannot really be applied to the typical clinical situation where a patient needs to undergo imaging to characterize an indeterminate renal mass.

Recently, Ozülker et al¹⁴ prospectively examined 18 patients with “suspicious” renal masses detected on CT, MRI, or ultrasound (US) with sizes ranging from 2 to 17 cm. FDG-PET/CT was performed within 4 weeks of CT/MRI/US. All patients underwent nephrectomy or surgical resection of the renal mass within the 2 weeks after completion of imaging studies. A renal mass was classified malignant when its intensity on PET was greater than the intensity of uptake in the renal parenchyma and it was distinct from the physiological excretion in the collecting system. Delayed images were acquired in 10 patients at 115 ± 5 minutes and compared with early images at 55 ± 5 minutes. Histology was used as gold standard after surgical resection of the mass, which was within 2 weeks after completion of imaging studies. There were 15 patients with RCC (14 ccRCC and 1 papillary RCC) and 3 with benign findings (2 renal cysts and 1 oncocytoma). PET was true positive for malignancy in 7 of the 15 lesions and false negative in 8 lesions (sensitivity 46.6%), true negative in 2 lesions (renal cysts) and false-positive in 1 lesion (oncocytoma) (specificity 66.6%). One of the cancers missed on PET was a papillary variant, and 7 of them were clear cell; the most variant of RCC was missed in 50% of the cases on FDG-PET. The median diameter and Furhrman grade of FDG-positive malignant lesions were higher than in FDG-negative malignant lesions (P < 0.05 in both cases). In 10 patients with delayed images, the delayed maximum standard uptake value (SUV_max) was 4.1 ± 3.0 versus early SUV_max of 4.2 ± 2.7, and the delayed SUV_mean was 2.1 ± 1.0 versus early SUV_mean of 2.1 ± 0.7 with no statistically significant in either case. Although the study by Ozülker et al is prospective and uses histology as gold standard, it again characterizes masses that are destined for resection and cannot be really called indeterminate. However, a useful conclusion of these data is that probably about half of ccRCC tumors are positive on FDG-PET; a conclusion that may help to draw prognostic information for this cancer as we will discuss later.

¹²⁴I-cG250 (Immuno) PET

The ccRCC represents only about half of the tumors enhancing on contrasted CT.^15,16 At the same time, about 90% of patients who present with, or later develop, metastatic renal cancer have the clear cell histological subtype.^15,17 Therefore, identifying ccRCC among the masses enhancing on CT is very important. Biopsy of renal masses is not done routinely because it is technically challenging and might cause seeding of tumor cells along the path of the biopsy needle.

¹²⁴I-cG250 is chimeric girentuximab labeled with ¹²⁴I (Redectane). cG250 functions as an epitope of CAIX.¹⁵ CAIX protein is a transmembrane enzyme involved in cellular pH regulation and appears to play a role in the regulation of cell proliferation in response to hypoxic conditions and may also be involved in oncogenesis and tumor progression.^18,19 CAIX is expressed in several cancers, such as renal, ovarian, colorectal, lung, brain, and bladder cancer, but absent in most normal tissues with the exception of epithelial cells of the stomach, small bowel, and bile duct.^18,20 Although the expression of CAIX varies in most cancers, it is almost universally expressed in ccRCC. In a study by Genega et al,²¹ all 184 cases of ccRCC expressed CAIX on immunohistochemistry while expression was high (>85% of the tumor cells) in 131 cases (71.2%) and low in the remainder 53 cases (28.8%). By contrast, CAIX expression was low or absent in the vast majority of non-ccRCC cancers.

CAIX as a potential tumor marker was first described by Oosterwijk et al.^18,22 At that time, the identity of CAIX was not known and so it was named after the antibody Grawitz250 (G250), with which it had been identified.¹⁸ In their original report, Oosterwijk et al describe G250 as a monoclonal antibody that recognizes an antigen preferentially expressed on cell membranes in RCC and not expressed in normal proximal tubular epithelium, from where RCC originates. In their report, G250 antibody reacted with 46 of 47 primary RCC tumors, with 7 of 8 RCC metastases and with a few other malignant tumors. In all tumors other than RCC that were studied with G250, the staining was much weaker than in RCC. Remarkably, in 43 of 47 primary RCC tumors and 5 of 8 RCC metastases, >50% of the cells were positive for G250, and staining was homogeneous throughout the tumor. Among many normal tissues studied, G250 staining was positive only in stomach, small bowel, and bile duct. Furthermore, Oosterwijk et al²² visualized human RCC xenografts implanted in nude mice with ^99mtechnetium-labeled G250.

G250 and its chimeric form, cG250, had a long journey and have been studied as an imaging agent. Most recently, Divgi et al studied 26 patients with renal masses suspicious for malignancy scheduled to undergo resection using cG250 labeled with ¹²⁴I. Each patient was injected with 5 mCi/10 mg of ¹²⁴I-cG250 over 20 minutes in this open-label pilot study followed by PET of the abdomen 6–8 days (median 7) later. The images were visually evaluated as positive or negative for malignancy by 2 nuclear medicine physicians. One patient was excluded from the final analysis because of administration of immunologically inactive ¹²⁴I-cG250. No agent-related immediate or delayed toxic effects were reported. After surgery, resected tumors were histopathologically classified as ccRCC or otherwise. Of the 16 patients with a pathological diagnosis of ccRCC who received immunologically active antibody, 15 were identified correctly with PET, resulting in a sensitivity of 94% (95% confidence interval (CI): 70%-100%). One patient was negative on PET, likely due to extensive necrosis, demonstrated on histology. All 9 patients without ccRCC were identified correctly by PET, resulting in a specificity of 100% (66%-100%). Positive and negative predictive values were 100% (78%-100%) and 90% (55%-100%), respectively. Fifteen of 16 clear cell renal carcinomas were positive for CAIX by immunohistochemistry. Quantitative measurement of uptake revealed that the uptake in ccRCC was 5.3–24.1 times higher than in normal renal cortex, whereas in non-ccRCC tumors the highest measured uptake was at most just 1.4 times higher than the normal cortex¹⁵ (Fig. 1). This study was followed by a multicenter study. The results are expected to be published soon.

Figure 1.

¹²⁴I-cG250 positron emission tomography (PET) images from a patient with biopsy-proven clear cell renal cancer: representative coronal computed tomography (CT) (A), ¹²⁴I-cG250 PET (B), fused image PET/CT (C) and transaxial CT (D), and ¹²⁴I-cG250 PET (E) images. Reprinted from the Lancet, Divgi et al,¹⁵ with permission from Elsevier. (Color version of figure is available online.)

As an aside, one should note that ¹²⁴I has a physical half-life of 4 days, and ¹²⁴I decays 75% by electron capture and only 25% by positron emission. Although the long half-life is advantageous, permitting centralized production of ¹²⁴I-cG250 and delivery to remote locations, the low positron abundance necessitates long acquisition times. Furthermore, the high photon energy of ¹²⁴I results in a high effective radiation, limiting the amount of radiotracer that should be injected.¹⁵

PET in Staging and Restaging of RCC

FDG-PET

About 30% of patients diagnosed with renal cancer have evidence of metastatic disease at the time of diagnosis. Additionally, 30% or more of those undergoing nephrectomy for localized disease later develop metastases.²³ Lung, bone, liver, and brain are the most common sites of distant metastases in RCC. Traditionally, contrasted CT of the chest, abdomen, and pelvis is performed to stage RCC patient with RCC. The reported sensitivity of FDG-PET for detecting RCC metastases is higher than for the primary RCC. Kang et al²⁴ studied 66 patients to assess the value of FDG-PET for primary RCC, metastatic disease, and local recurrence. PET detected 66.9% of distant metastases and 75% of retroperitoneal metastases/local recurrences versus only 60% of the primary RCC tumors. On a patient basis, there were only 6 out of 52 patients (11.6%) in whom all metastatic lesions were missed on PET. In the remaining 46 patients (88.4%), PET detected at least some of the distant metastases, but failed to identify others. The authors attributed this to variable intensities of different metastatic lesions on FDG-PET. Majhail et al²⁵ studied 24 patients with histologically proven ccRCC to assess the value of FDG-PET for detecting distant metastases. In total, 36 metastatic sites were identified based on CT and MRI and were further evaluated by biopsy and histology. There were histologically documented metastases in 33 sites in 21 patients. Overall sensitivity, specificity, and positive predictive value of FDG-PET for the detection of distant metastases were 63.6%, 100%, and 100%, respectively. The mean size of distant metastases in patients with true-positive PET finding was 2.2 cm (95% CI: 1.7–2.6 cm) compared with 1.0 cm in patients with false-negative PET finding (95% CI: 0.7–1.4 cm; P = 0.001). The sensitivities reported by Kang et al and Majhail et al for FDG-PET in detecting RCC metastases are higher than in the literature reported sensitivity of about 50% for detecting the primary RCC by FDG-PET, although the site of the primary RCC is usually larger than the metastases. This indicates the probably different biology and possibly higher expression of Glut and/or hexokinase in the metastatic sites compared with the primary RCC. Kurata et al²⁶ reported significantly higher expression of Glut3 and Glut5 in the hepatic metastases of lung cancer than in the primary site of the disease. There are case reports available from our group and others, demonstrating that the variability in FDG uptake of different metastatic sites from the same tumor is related to differences in Glut1 expression.^27,28 At the same time, it should also be noted that both of the publications by Kang et al and Majhail et al were based on data of almost a decade ago from PET-alone machines, and one should expect higher sensitivity and specificity on the current combined machines with improved PET resolution and concurrent CT acquisition (PET/CT).

Nakatani et al²⁹ studied the value of FDG-PET to detect recurrence disease in 23 postsurgical RCC patients. Overall, the sensitivity, specificity, and diagnostic accuracy of FDG-PET for detecting recurrent malignancy were 81%, 71%, and 79%, respectively. In ccRCC, the sensitivity was 75%. PET correctly detected all cases of intra-abdominal (local recurrence, lymph nodes, and adrenal glands) and bone recurrence. There was a trend for better 5-year survival in PET-negative patients compared with PET-positive patients of 83% versus 46%, respectively (P = 0.17). This underlines the fact that FDG-PET has probably a prognostic value in RCC as we discuss next.

cG250 PET

It is conceivable that ¹²⁴I-cG250 PET will play a role in staging and restaging of ccRCC. But one has to wait for the anticipated reports.

PET in Predicting and Monitoring Therapy Response in RCC

Tumors are often classified based on tissue biomarkers. Tissue biomarkers have prognostic value and can guide treatment and predict outcome of the disease. During the course of treatment, tissue biomarkers may be used to monitor the effectiveness of treatment. However, there are problems associated with using tissue biomarkers. First and foremost, tissue biomarkers necessitate biopsy, an invasive procedure, which needless to say is prudent to be avoided as often as possible. At the same time, biopsy often cannot produce enough tissue from the representative area of the tumor for comprehensive analysis. Even after obtaining representative tissue samples, there are several other parameters that are known to potentially alter mRNA and protein expression levels within the biopsied or resected tissue and consequently bias the results of molecular analyses. These factors include, but not limited to type and duration of anesthesia, presence and duration of tissue ischemia after clamping or blood supply, tissue storage conditions (ie, medium and temperature). In addition, the use of laparoscopic procedures has introduced new variables, including the use of distension medium (eg, carbon dioxide), causing a higher than normal intra-abdominal pressure that can potentially affect baroreceptors and modulate molecular pathways.⁹ At the same time, it is generally accepted that the prognostic impact of the features of primary tumor disappears once the tumor spreads and becomes metastatic and that there are changes that tumor cells are likely to undergo over time.³⁰ Therefore, exact knowledge of various markers and underlying pathways that are present in the metastatic disease at the time of treatment is paramount for proper selection of targeted agents to increase the likelihood of therapy success. For these and other reasons, using imaging biomarker strategy as an alternative to tissue biomarker is very attractive and is gaining momentum.

The majority of ccRCC patients have metastatic disease at some point during the course of the disease, and therefore, in need for systemic therapy. The paradigm of treatment for metastatic ccRCC has shifted away from immune-based therapy in the recent years, and targeted drugs have been developed to halt tumor growth by aiming specific pathways. The fact that these drugs are administered in ccRCC most often as single agent makes PET imaging with a radiotracer that visualizes steps in the metabolic pathway of the targeted drug an attractive biomarker for predicting and monitoring the effect of the drug.

FDG-PET

Two major groups of targeted drugs that are currently approved for use in metastatic RCC are multikinase inhibitors and mTOR inhibitors. Sorafenib and sunitinib are 2 representative multikinase inhibitors, inhibiting the receptor tyrosine kinase VEGF receptor 2 and the platelet-derived growth factor receptor β in the endothelial cells and pericytes, respectively.³¹ Because expression of Glut is a downstream product of HIF transcriptional activity, it is conceivable that the intensity of FDG uptake on PET may be reflective of the magnitude of the entire pathway. This means that FDG-PET with its variable intensity in ccRCC may reflect the variable strength of the HIF signaling pathways and expression of its downstream products and be predictive of the effect of the inhibitors of this pathway. Revheim et al³² studied the utility of pretherapy SUV_max on FDG-PET in predicting progression-free survival (PFS) in 14 patients (12 ccRCC patients) with metastatic RCC undergoing therapy with sunitinib. Using an arbitrary cutoff of 5 for SUV_max for the most intense lesions in each patient, they demonstrated a longer PFS in patients with an SUV_max of <5 compared with those with SUV_max of >5.

The largest study addressing the value of FDG-PET in assessing the response to a targeted drug in RCC was conducted by Kayani et al.³³ They studied prospectively 44 treatment-naive ccRCC patients only, which were part of a phase II multicenter trial of sunitinib. FDG-PET scans were performed before treatment and at 4 and 16 weeks. In each patient, the most intense lesion (primary or metastatic) was used as the index lesion as long as it had an SUV of >2.5. Metabolic response was defined as a decrease of >20% in SUV. Metabolic disease progression was defined as an increase of ≥20% in SUV or development of new metastatic sites. After 4 weeks of sunitinib, a metabolic response occurred in 24 (57%) patients, but this did not correlate with PFS (hazard ratio [HR] for responders = 0.87 [95% CI: 0.40–1.99]) or overall survival (OS) (HR for responders = 0.80 [95% CI: 0.34–1.85]). After 16 weeks of treatment, 12 patients (28%) had metabolic disease progression, which correlated with a decreased OS and PFS (HR: 5.96 [95% CI: 2.43–19.02] and 12.13 [95% CI: 3.72–46.45]), respectively. Ten out of the 12 patients with disease progression at 16 weeks had a response to therapy at 4 week.³³ One simple conclusion of this study is the fact that in this disease with overall very poor response rate to therapy, it would be probably more feasible to predict disease progression and not response and try to exclude some of the many nonresponders as early as possible after treatment starts. At the same time, the fact that 10 out of the 12 patients with disease progression at 16 weeks had a response to therapy at 4 weeks might be explained by the fact that the proposed “minimum” of 20% change in SUV to distinguish true change in tumor metabolism from interscan variability might be too small for differentiating responders from nonresponders when sunitinib (or perhaps also other targeted drugs) is used; in which case, changes in tumor metabolism occur rather slowly.³⁴ Therefore, a larger change from the baseline SUV may be needed to distinguish responders from nonresponders. Another result of this publication was that in multivariate analysis, a high SUV_max at baseline and an increased number of PET-positive lesions correlated with shorter OS (HR: 3.30 [95% CI: 1.36–8.45] and 3.67 [95% CI: 1.43–9.39], respectively). Association between higher SUV and shorter OS again points to the fact that sunitinib effect and FDG uptake share the same signaling pathway, and that, perhaps, because of the tonic signaling of the HIF pathway to maintain the endothelial cells, tumors with higher pretherapy FDG uptake respond less well to VEGF receptor targeted therapy with sunitinib compared with those with lower uptake because the inhibitor is less likely to overcome the strength of the signal. Predicting the outcome of therapy based on the pretherapy parameters is very attractive because it could potentially avoid unnecessary treatment in some patients all together. The findings by Kayani et al are along our experience that ccRCC tumors with a lower pretherapy uptake on FDG-PET demonstrate a larger size decrease on CT after treatment with tyrosine kinase inhibitors³⁵ (Fig. 2).

Figure 2.

Representative images from patients with clear cell cancer and targeted therapy with sorafenib. Low tumor uptake on fluorodeoxyglucose PET before start of therapy tends to predict a larger size decrease on CT after treatment. The tumor with the pretherapy maximum standard uptake value of 3.3 (A) had a size decrease of 18%, whereas the tumor with the baseline maximum standard uptake value of 13.1 (B) had a size decrease of only 3%.(Color version of figure is available online.)

¹²⁴I-cG250PET

Most ccRCC tumors express abundantly CAIX, which is a downstream pathway product of HIF transcriptional activation. Oosterwijk-Wakka et al³⁶ demonstrated that the uptake of ¹²⁵I-cG250 in human RCC implanted in mice decreased after 7 days of treatment with tyrosine kinase inhibitor, whereas there was a rebound of ¹²⁵I-cG250 uptake as well as robust neovascularization when the treatment was stopped. It is conceivable that the similar results could be achieved in human with ¹²⁴I-cG250 PET but there are no publications available in this regard at this time. At the same time, there is some evidence that CAIX may be a predictive factor for response of ccRCC to targeted drugs. Choueiri et al³⁷ investigated the utility of CAIX expression in 94 metastatic ccRCC for predicting response to first-line therapy with sorafenib and sunitinib with tumor shrinkage as the primary endpoint. The mean shrinkage was −17% versus −25% for sunitinib-treated patients with high versus low tumor CAIX expression, compared with −13% versus +9% for sorafenib-treated patients (P interaction = 0.05). The authors suggested that CAIX might be a predictive biomarker for response to sorafenib treatment. CAIX has also been reported as an independent prognostic factor in ccRCC. Bui et al¹⁹ reported that low CAIX expression in primary ccRCC was an adverse predictor of survival in a large series of 321 ccRCC patients. In the multivariate analysis, CAIX expression remained an independent prognostic parameter. Although the originally reported finding of CAIX by Bui et al as an independent prognostic factor has been confirmed by others,^38,39 Leibovich et al⁴⁰ found in a series of 730 ccRCC patients that although low CAIX expression was associated with increased risk of RCC death, this was no longer the case after adjusting for nuclear grade or coagulative tumor necrosis. Despite some discrepancy in results of various studies, which could partly be related to the number of metastatic cancer patients included in each study and variable tissue examination methods, it appears that CAIX has a value as prognosticator and predictor of response to targeted therapy in ccRCC. It should be noted that CAIX appears to have the opposite prognostic value in ccRCC than in other malignancies, such as breast, head, and neck cancer, where its high expression is associated with worse prognosis.⁴¹ Although the reason of this discrepancy is not clear, one should be reminded that the mechanism behind the expression of CAIX is very different in ccRCC than in other malignancies. In ccRCC, CAIX expression is associated with the absence of VHL protein and unrelated to hypoxia, whereas its expression in other malignancies is a result of the hypoxic environment.⁴¹ It has been suggested that low CAIX expression in ccRCC might be related to tumor dedifferentiation and aggressiveness.²⁰ In any case, we look forward to seeing what a role immuno-PET with ¹²⁴I-cG250 may play in predicting the response of targeted drugs in ccRCC.

¹⁸F-FLT PET

Wong et al⁴² studied the uptake of FLT and FDG in 27 preoperative patients with newly diagnosed primary renal cell cancers (16 clear cell, 5 papillary, 1 chromophobe, 1 sarcomatoid, 1 transitional cell, and 3 benign tumors) and compared the tumor uptake on PET with tumor proliferation rate measured as Ki-67 expression on immunohistochemistry on the resected tumor from partial or total nephrectomy. The grades of RCC tumors ranged from 1 to 4. The mean Ki-67 ± SD was 13.3% ± 9.2(range 2.2%–36.3%). FLT uptake was detected on PET in all tumors. The mean FLT SUV_max ± SD was 2.53 ± 1.26 and correlated strongly with expression of Ki-67 (r = 0.624, P = 0.0008, Pearson correlation). The mean SUV_max ± SD for FDG uptake in the same tumors was 2.60 ± 1.08 and correlated strongly with Ki-67 expression (r = 0.75, P < 0.001). This was published in the form of an abstract and further details, such as the size of the examined tumors and FLT uptake in the ccRCC category, were not given. But it should be noted that the size of tumors must have been rather small in these surgical candidates, some of whom underwent partial nephrectomy. Further studies with a broad size range of primary tumors and metastatic disease are desired to further assess the uptake of FLT in ccRCC, the largest and most aggressive variant of RCC. It is conceivable that FLT PET could play an important role in predicting and monitoring the response to targeted drugs in ccRCC. However, one has to keep in mind that 2 organs, which are often site of metastatic disease from RCC, liver and bone, cannot be reliably evaluated by FLT because of physiological uptake.

Liu et al used FLT PET to characterize and quantify changes in tumor proliferation during sunitinib exposure and temporary withdrawal and to explore pharmacodynamics changes that may yield insight into predicting treatment response. They studied 7 patients with metastatic RCC and 9 patients with a very diverse group of 7 other malignancies. Patients were on 4/2 (4 weeks of sunitinib treatment followed by a 2-week break) or 2/1 (2 weeks of treatment followed by a 1-week break). FLT PET scans were performed at the baseline, at the end of treatment period and at the end of the break period. When all tumors and treatment schedules were combined, there was a statistically significant decrease in SUV_mean and SUV_max at the end of treatment period and a statistically significant increase (flare) in SUV_mean and SUV_max at the end of the break period. Also across all patients, there were statistically significant increases and decreases in the plasma VEGF and sunitinib levels during sunitinib treatment and withdrawal period, respectively. After adjusting for sunitinib concentration, across all patients, VEGF change was found to be negatively correlated with changes of SUV_mean and SUV_max during sunitinib treatment with partial Spearman rank correlations of −0.69 (P = 0.0041) and −0.71 (P = 0.0030), respectively. Similarly, VEGF change was found to be negatively correlated with changes of both SUV_mean and SUV_max during sunitinib withdrawal after adjusting for sunitinib concentration across all patients (rps = −0.82, P = 0.0002; rps = −0.85, P = 0.0001). Multivariate analysis, involving both changes in VEGF and sunitinib concentration as predictors for changes in SUV_mean, showed that changes in VEGF independently predicted change in SUV_mean during the sunitinib treatment period (P = 0.003). Furthermore, the association between changes in pharmacodynamic parameters and clinical response by RECIST was evaluated. When treatment groups were combined, nonresponders by RECIST (n = 9) had statistically significant median increases in SUV_mean (+29%, range: −5%-+277%; P = 0.012) and SUV_max (+36%, range: −11%-+221%; P = 0.039) during sunitinib withdrawal. Responders (n = 5) appeared to have a trend toward unchanged SUV_mean (+0.00%, range: −14%-+8%; P = 0.648) and SUV_max (+2.6%, range: −33%-+31%; P = 0.963). This suggests that a large flare might be an early sign of treatment failure. Overall, the data by Liu et al⁴³ underline the dynamic nature of FLT uptake in tumor and its association with pharmacokinetic parameters. Although data specifically concerning RCC were not given, FLT has the potential to play an important role, specifically in drug development in this malignancy, and further research is desired.

¹¹¹In-Bevacizumab PET

In ccRCC, accumulation of HIF-1α, which is again due to VHL defect and independent of hypoxia, leads to upregulation of downstream kinases, ultimately increasing the production of products, such as VEGF-A. Therefore, molecular imaging of VEGF-A in ccRCC might help to monitor the effects of tyrosine kinase inhibitors, such as sunitinib and sorafenib.¹⁰

Bevacizumab is a humanized monoclonal antibody directed against all VEGF-A isoforms, inhibiting angiogenesis by preventing VEGF-A from binding to and activating its receptors. VEGF-A is the best-characterized member of the VEGF family and is considered the predominant and most critical regulator of neovascularization in various tumor types.¹⁰ Bevacizumab is currently approved in the United States by the Food and Drug Administration as a targeted drug for treatment of several cancers; among others, for metastatic RCC in combination with interferon alfa. Desar et al evaluated the potential of radiolabeled bevacizumab as an imaging biomarker. They studied ¹¹¹In-bevacizumab in 8 patients with primary ccRCC and 1 patient with primary papillary RCC who underwent neoadjuvant therapy with sorafenib for 4 weeks, followed by nephrectomy. These patients underwent ¹¹¹In-bevacizumab whole-body imaging before the start and after the completion of sorafenib. Five patients without neoadjuvant treatment underwent ¹¹¹In-bevacizumab before nephrectomy and served as control. In each patient, whole-body images were acquired 7 days after intravenous injection of 100 MBq (2.7 mCi) ¹¹¹In-bevacizumab. In every patient, there was accumulation of radiotracer in the regions with expected viable tumor. In the ccRCC patients, a mean decrease of 60.5% of ¹¹¹In-bevacizumab uptake in the primary tumors was observed (median, −46.2%; range, +1.5%-−90.1%; P = 0.011) after sorafenib treatment. Also decrease of radiotracer uptake was observed in metastases; a lung metastasis measuring >1 cm had a decrease of >95% of uptake. The pattern of uptake of the ¹¹¹In–labeled antibody in the tumor slices corresponded well with macroscopic discernible vital and necrotic areas in the tumor, with a relatively high uptake in the vital areas and low uptake in the necrotic areas (Fig. 3). In the only patient with papillary RCC, there was relatively low uptake compared with ccRCC tumors at baseline, which did not change significantly after sorafenib therapy.¹⁰ These data indicate that ¹¹¹In-bevacizumab is able to depict ccRCC and that neoadjuvant treatment with sorafenib significantly reduces its accumulation in this tumor. Further research, perhaps with ¹²⁴I–labeled antibody and quantification on PET, is needed to evaluate the role of radiolabeled bevacizumab as an imaging biomarker to stratify patient to antiangiogenic therapy and for therapy monitoring.

Figure 3.

Anterior and posterior ¹¹¹In-bevacizumab images of a patient with left kidney clear cell cancer baseline (A) and after 4 weeks of treatment with sorafenib (B). Decrease of ¹¹¹In-bevacizumab uptake, more enhanced in central parts of tumor, is shown (arrows). Reprinted by permission of the Society of Nuclear Medicine from Desar et al.¹⁰

¹⁸F-Fluoromisonidazole (FMISO) PET

Lawrentschuk et al⁴⁴ studied 11 primary RCC tumors with FMISO PET. Although there was a trend, there was no statistically significant difference in SUV_max when comparing tumors with normal cortex of the contralateral kidney (mean ± SD: 1.3 ± 0.15 vs 1.1 ± 0.22; P = 0.14). Polarographic measurement of pO₂ in 3 tumors with clear cell papillary and chromophobe histology revealed 9.7, 5.2, and 24.9 compared with respective normal renal tissue values of 47.6, 14.9, and 39.5.

Hugonnet et al⁴⁵ use FMISO in 53 patients with metastatic RCC at baseline and 1 month after sunitinib treatment. Lesions were considered hypoxic when their maximal SUV was above mean blood value + 2SD. Seventy-two of the lesions were hypoxic at the baseline with a geometrical mean of the SUV_max of 1.86 (95% CI: 1.74–2.00), whereas 111 lesions were not. Seventy-five percent of patients with hypoxic metastases were free of progressive disease at 4.8 months (95% CI: 2.99–11.83), compared with 11.3 months (95% CI: 3.08–36.9) for other patients (P = 0.02), whereas OS was not significantly different. Changes in tumor hypoxia at 1 month after therapy were not related to PFS or OS. The conclusion of the authors was that hypoxia in metastatic RCC as assessed by FMISO PET was less frequent and less pronounced than initially suspected, and the authors questioned if FMISO PET performed earlier than a month of initiation of sunitinib therapy could detect more acute hypoxia caused by the effect of the drug on tumor perfusion.

The results by Lawrentschuk et al and Hugonnet et al indicate that hypoxia in RCC is probably less severe than one might assume. One should bear in mind that HIF and its downstream pathway products are increased in ccRCC because of defective VHL gene regulation and should not be taken as evidence for hypoxic condition as is the case in other malignancies. Rather, this condition represents a “pseudohypoxic” state, in which, hypoxia response pathway is on and constitutively activated, but often times in the presence of abundant oxygen availability.

¹¹C-Acetate (AC)

The data on ¹¹C-AC PET in RCC are limited and contradictory. Kotzerke et al⁴⁶ studied 18 RCC tumors with AC PET. AC uptake in renal tumors was less than in the normal renal cortex in 11 cases, equal to the normal cortex in 5 cases and higher than the normal renal cortex in just 2 cases. Both tumors with higher AC uptake than in normal cortex (PET positive) were oncocytomas. In 2009, Oyama et al⁴⁷ studied 22 renal tumors, 20 RCC, and 2 complicated renal cysts in 20 patients. There were 19 ccRCC of which 13 (69%) were positive on PET. The only papillary cell carcinoma was also PET positive, whereas both benign tumors were PET negative. It should be noted that Kotzerke et al considered tumors as PET positive when they were “intense;”, that is, grade 4 on a scale from 0 (photopenic) to 4, whereas the article by Omaya et al does not give any specific information on PET interpretation. Although various thresholds for PET positivity could account for some of the significant discrepancy between the results of these 2 studies, they probably cannot completely explain it. Further studies could help to assess the usefulness of AC PET for RCC.

Conclusions

It is conceivable to believe that PET with the multitude of radiotracers now on the market and many in the pipeline would complement and, maybe one day, replaces tissue biomarkers in renal carcinomas. PET as an imaging biomarker has the potential to provide prognostic and predictive information and helps to personalize the care of patients suffering from this devastating disease. An area of research we call “PETology.”