Advertisement

Diagnostic Applications of Nuclear Medicine: Gastrointestinal Stromal Tumors

  • Christopher G. Sakellis
  • Heather A. Jacene
  • Annick D. Van den AbbeeleEmail author
Living reference work entry

Abstract

Gastrointestinal stromal tumor (GIST), although rare, is the most common mesenchymal gastrointestinal neoplasm. This chapter will review the epidemiology, environmental factors, genetic predisposition, and underlying biomolecular changes of the disease. The staging of GIST as well as the roles of conventional diagnostic imaging and nuclear imaging in this staging will be reviewed. Finally, the efficacy of these modalities in assessing response to the various treatments described in the chapter and in the long-term surveillance for disease recurrence will be addressed.

Keywords

Gastrointestinal stromal tumor (GIST) Conventional diagnostic imaging [18F]FDG-PET/CT Staging Tyrosine kinase inhibitors Assessment of therapeutic response Surveillance 

Glossary

[18F]FDG

2-deoxy-2-[18F]fluoro-d-glucose

ADC

Apparent diffusion coefficient

AFIP

Armed Forces Institute of Pathology

AJCC

American Joint Committee on Cancer

BRAF

Gene encoding for the B-Raf protein, a serine/threonine-protein kinase; the gene is also known as the proto-oncogene B-Raf and v-Raf murine sarcoma viral oncogene homolog B

CSF-1R

Colony-stimulating factor 1R

CT

X-ray computed tomography

DECT

Dual-energy CT

EORTC

European Organization for Research and Treatment of Cancer

EUS

Endoscopic ultrasound

FDA

United States Food and Drug Administration

GIST

Gastrointestinal stromal tumor

GLUT4

Glucose transporter 4

HPF

High-power fields

HU

Hounsfield units

IRA

Iodine-related attenuation

KIT

A proto-oncogene encoding for tyrosine-protein kinase Kit (or CD117), also known as mast/stem cell growth factor receptor (SCFR)

M

Metastasis status according to the AJCC/UICC TNM staging system

MRI

Magnetic resonance imaging

N

Lymph node status according to the AJCC/UICC TNM staging system

NCCN

National Comprehensive Cancer Network

NF1

Neurofibromatosis type 1

PDGFR

Platelet-derived growth factor receptor

PDGFRA

Platelet-derived growth factor receptor alpha

PET

Positron emission tomography

PET/CT

Positron emission tomography/Computed tomography

RECIST

Response evaluation criteria in solid tumors

RET

A proto-oncogene encoding for a receptor tyrosine kinase for extracellular signalling molecules (from “rearranged during transfection”)

RTK

Receptor tyrosine kinases

SCF

Stem cell factor

SDH

Succinate dehydrogenase

SURE

Sunitinib alternating with regorafenib

SUVmax

Standardized uptake value at point of maximum

SUV

Standardized uptake value

TKI

Tyrosine kinase inhibitor

TNM

AJCC/UICC staging system based on parameters “T” (tumor status), “N” (lymph node status) and “M” (distant metastasis status)

UICC

International Union Against Cancer (Union Internationale Contre le Cancer)

T

Tumor status according to the AJCC/UICC TNM staging system

TEK

Receptor tyrosine kinase

VEGFR

Vascular endothelial growth factor receptor

Epidemiology

Gastrointestinal stromal tumor (GIST) is the most common mesenchymal gastrointestinal neoplasm but is still a rare tumor, only accounting for about 1% of all gastrointestinal neoplasms [1, 2]. The incidence of the tumor in the United States is estimated to be 4,000–6,000 new cases per year [3].

GISTs are distinct from other mesenchymal gastrointestinal tumors like liposarcomas or leiomyosarcomas, as they are derived from cells in the myenteric plexus of the lamina propria within the gastrointestinal wall [4]. They can arise anywhere throughout the gastrointestinal tract, most commonly in the stomach (60%) and jejunum/ileum (30%), followed by the duodenum, rectum, colon, and esophagus in descending order [5]. Rarely, GISTs have been found to arise in the mesentery or retroperitoneum.

The majority of GISTs are sporadic, and in these cases, men and women are affected equally. There is no racial preference, and they rarely arise before 40 years of age, with a mean age at diagnosis in the seventh decade of life [6].

Environmental Factors and Genetic Predisposition

At this time, there are no known environmental or lifestyle-related causes of GIST. However, almost all GISTs have been found to be associated with sporadic activating mutations which promote tumor survival and growth.

Approximately 85% of GISTs have oncogenic gain-of-function mutations in one of two receptor tyrosine kinases (RTK): KIT or PDGFRA (platelet-derived growth factor receptor alpha) which play a central role in the pathogenesis of GIST [7, 8, 9]. KIT encodes for CD117, a transmembrane receptor tyrosine kinase protein. The most common mutation in KIT in GIST involves exon 11 (about 70% of cases), while other mutations have been found in exons 9 (10–20% of cases), 13, and 17 [10]. Various types of mutations have been described, most commonly in-frame deletions involving exon 11 and duplications in exon 9.

The activating mutations involving the PDGFRA gene which is homologous to KIT and also encodes a transmembrane receptor tyrosine kinase have been found in 5–10% of GISTs [8]. Single nucleotide substitutions are the most common mutations described in these PDGFRA-positive tumors. KIT-negative GISTs are more likely to have PDGFRA mutations than KIT-positive GISTs [11]. Approximately 10–15% of GISTs are considered wild type and are both KIT- and PDGFRA-negative [10].

Although rare, inherited predispositions have been found to account for 5% or less of all GISTs. Patients with neurofibromatosis type 1 (NF1) have a high incidence of GISTs, which are typically multiple, arise in the small intestine, and are both KIT- and PDGFRA-negative [12]. Overexpression of KIT can occur in these wild-type NF1-related GISTs, though the mechanism is unclear.

Pediatric GIST is rare, tends to arise in the setting of heritable syndromes, has a female predilection, and commonly presents with chronic gastrointestinal bleeding and multifocal gastric involvement. Unlike adult GISTs, pediatric GISTs lack KIT or PDGFRA mutations, commonly metastasize to the lymph nodes, and have a more indolent course. The Carney-Stratakis syndrome results from germline mutations in succinate dehydrogenase (SDH) genes and presents in children who go on to develop GISTs and paragangliomas [13]. Primary familial GIST syndrome is a very rare autosomal dominant phenomenon and involves a germline mutation in KIT with high penetrance. These patients tend to be younger, develop multiple tumors in the stomach and/or small bowel, and may display extra-gastrointestinal traits such as skin hyperpigmentation [14].

A subset (3–13%) of wild-type GISTs has also been found to possess mutations in the serine-threonine kinase BRAF [15].

Underlying Biomolecular Changes

CD117 , a transmembrane receptor tyrosine kinase protein, is encoded for by the KIT gene and functions as a receptor for stem cell factor (SCF). In normal cells, CD117 is enzymatically inactive until it binds SCF , whereby it initiates a signaling cascade within the cell. When there is a “gain-of-function” mutation in the KIT gene, as seen in most GISTs, the mutated CD117 protein remains activated without the presence of its ligand SCF, and thus the signaling cascade continues unabated. The most common mutations of the KIT gene which occur in exon 11 specifically affect the intracellular juxtamembrane portion of CD117, which plays a role in inhibiting signal transduction in the absence of SCF [16]. Mutations in other exons of the KIT gene as well as mutations seen in GIST in the PDGFRA gene all result in constitutive kinase activation and unchecked cell signaling, which ultimately affects cell proliferation, apoptosis, chemotaxis, and metabolism, resulting in tumorigenesis.

Staging and Prognostic Stratification

The TNM (tumor-node-metastasis) staging system developed by the American Joint Committee on Cancer (AJCC) and the International Union Against Cancer (UICC) is used for GIST. Although the T, N, and M definitions are similar for all tumors, there are separate stage groups based on the location of the primary tumor. Primary gastric and omental GISTs are grouped together in one scheme, while primary small bowel, esophageal, colorectal, and peritoneal GISTs are grouped in another. Based on the data of two large retrospective Armed Forces Institute of Pathology (AFIP) studies of GISTs with long-term follow-up, tumor-related mortality of small bowel GISTs was more than twice that of gastric GISTs [17, 18].

Tumor (T) staging includes an assessment of both tumor size and mitotic rate. As most GISTs are roughly ovoid, the greatest tumor diameter is relatively easy to assess. The size thresholds in assigning a T designation are 2, 5, and 10 cm. Depth of tumor invasion is not utilized as most GISTs have transmural extension at presentation. The mitotic rate of GIST is evaluated pathologically by determining the number of mitoses observed per 50 high-power fields (HPF) on microscopy. A low mitotic rate is defined as 5 or fewer mitoses per 50 HPF, while a high mitotic rate is over 5.

Based on AFIP data, GISTs smaller than 2 cm in size with low mitotic rates were found to have very low risk of recurrence but were still considered to have metastatic potential. In both gastric and small bowel GISTs, less than 3% of tumors less than 10 cm and with fewer than 5 mitoses per 50 HPF metastasized, while over 80% of tumors larger than 10 cm and with 5 or more mitoses per HPF metastasized. The more intermediate-risk gastric GISTs (greater than 10 cm with low mitotic rate/less than 10 cm with high mitotic rate) had relatively low metastatic potential (10–15%), while the intermediate-risk small bowel GISTs with the same characteristics had a much higher metastatic rate (greater than 50%) [17, 18]. The National Comprehensive Cancer Network (NCCN) guidelines stratify the risk of malignancy of GISTs as high, moderate, low, or very low based on the site of primary tumor, size, and anatomic location [19].

Nodal (N) staging evaluates for the presence of regional lymph node metastases, which are rarely seen outside of rare pediatric familial cases. The presence of no regional nodal metastases or an unknown nodal status is designated as N0.

Distant metastatic disease (M1) in GIST is tumor involvement outside the main tumor mass, most commonly in the mesentery or liver within the abdomen. Extra-abdominal metastases are rare, but can occur in the bone, subcutaneous tissue, and lung. Multiple synchronous gastrointestinal GISTs, usually seen in the setting of hereditary syndromes, should not be considered as metastatic disease, as all could be separate primary tumors.

The histology of GIST may affect prognosis. The three major histopathologic types are spindle cell (70%), epithelioid type (20%), and mixed cellularity (10%). Early data suggested that 5-year recurrence-free survival was significantly higher with spindle cell histology compared to epithelioid or mixed histology [20]. However, subsequent studies implied that the degree of cellularity of the GIST was a more important prognostic factor than histologic type [21].

At this time, it is not clear whether the presence or absence of KIT mutations affects the prognosis of a particular GIST. Specific types of KIT mutations, including those in exon 9, have been shown to be associated with more aggressive tumor behavior [22]. Specific mutations are not factored into the current TNM staging system.

Conventional Diagnostic Imaging Staging

Contrast-enhanced CT is the current imaging of choice for the initial diagnosis and staging of GISTs . Although tissue sampling is required to definitively diagnose a GIST, CT not only characterizes the size and extent of the primary tumor but also evaluates for any peritoneal seeding and/or more distant metastatic disease. With the aid of oral and intravenous contrast, features associated with the primary tumor such as ulceration and fistulization with the gastrointestinal lumen can be characterized.

Typically, at the time of presentation, primary GISTs appear as smoothly contoured, heterogeneously enhancing masses. Larger tumors may display internal heterogeneity due to associated necrosis, cystic degeneration, or hemorrhage. It is important that multiphasic CT imaging with unenhanced, arterial phase, and portal venous phase images is performed when imaging the liver, as the surrounding hepatic parenchyma on delayed portal venous phase images performed alone may mask hypervascular hepatic metastases.

Though CT is better at globally evaluating the patient for metastatic disease, MRI has shown utility in evaluating disease, especially in specific fixed structures such as the liver or rectum. The higher soft-tissue contrast of MRI compared to CT often provides an advantage in determining the organ of origin of primary GISTs and in evaluating invasion or involvement of adjacent viscera and/or blood vessels.

Some MRI features of primary GISTs at baseline may be predictive of malignancy risk. The presence of intratumoral cystic change, which is more accurately assessed on MRI than CT, has been found to correlate with high-risk GISTs, as defined by the NCCN guidelines [23]. A similar correlation between the presence of low attenuation findings on CT and malignant potential has not been found, possibly because these findings can be more nonspecific on CT [24, 25]. On MRI, lower mean apparent diffusion coefficient (ADC) values have also been found to correlate with a higher malignant risk, possibly due to the higher cellularity (and associated diffusion restriction) of high-risk tumors [23].

Endoscopic ultrasound (EUS) has been considered for the diagnosis and management of very small GISTs (less than 2 cm). EUS-guided biopsy of the lesion can be performed at the time of evaluation. Certain high-risk EUS features of these very small GISTs have been described including irregular border, cystic spaces, ulceration, echogenic foci, and overall heterogeneity, which may prompt their surgical resection [26].

Nuclear Imaging Staging

Fluorodeoxyglucose-positron emission tomography/computed tomography ([18F]FDG-PET/CT) scans are now commonly used in the baseline evaluation of biopsy-proven GIST. CT and [18F]FDG-PET scans were initially shown to have comparable high sensitivity and positive predictive value in initial staging of GIST [27], while subsequent studies have demonstrated [18F]FDG-PET/CT is slightly more sensitive than CT alone [28]. While contrast-enhanced CT provides superior anatomic definition compared to the nondiagnostic CT portion of a PET/CT exam, the PET portion of a [18F]FDG-PET/CT may be more sensitive in detecting an unknown primary GIST and may more definitively characterize small findings on CT scan as sites of metastatic disease. [18F]FDG-PET/CT scans may also better evaluate patients at baseline who have an allergy to intravenous contrast.

[18F]FDG-PET allows for quantitation of tumor metabolism using measurements such as the maximum standardized uptake value (SUVmax). As PET has been shown to be superior in many ways to diagnostic CT in monitoring and quantifying therapeutic response to tyrosine kinase inhibitor (TKI) therapy (described below), especially early in the course of treatment, a baseline [18F]FDG-PET scan should be obtained before starting any TKI administration to assess the baseline tumor metabolic activity and extent of tumor burden.

Traditional Therapies and GIST

Complete surgical resection is the treatment of choice for any patient with surgically resectable GIST and no metastases and offers the best chance for long-term disease-free survival. The goal of surgery is complete resection of the primary tumor with an intact pseudocapsule and negative microscopic margins on pathology. The resected tumor needs to be handled with care during removal, as tumor rupture will negatively impact survival due to the risk that it results in new intra-abdominal microscopic seeding of tumor.

The current consensus is that conventional cytotoxic chemotherapy is not useful for managing GIST. Several trials evaluating chemotherapeutic regimens in advanced GIST have reported very low response rates (0–5%) [29, 30].

Radiotherapy has been reported in the past to show no advantage in disease-free survival in several studies. A retrospective study analyzing patients treated at Massachusetts General Hospital between 1973 and 1998 with incompletely resected GIST, some of whom were also treated with radiotherapy, concluded there was no support for its use [31]. GISTs are also commonly associated with mobile structures such as loops of small bowel and are more difficult to accurately target with radiotherapy.

Molecular Therapy and GIST

Until the early 2000s, the outcome for patients presenting with GISTs which were either advanced (surgically unresectable or marginally resectable) or who presented with metastatic disease was relatively bleak. After the discovery of the key role mutations in receptor tyrosine kinase genes, most notably KIT, played in GIST tumorogenesis, molecularly targeted agents known as tyrosine kinase inhibitors (TKIs) were developed to specifically treat GIST.

The first of these tyrosine kinase inhibitors used to treat GIST in clinical trials, and later the first to be approved by the FDA for treatment of advanced or metastatic GIST in 2002, was the small molecule imatinib mesylate. Imatinib inhibits the reaction that starts the cell signaling cascade of the KIT protein by competing with ATP for a binding site on its surface, thus inhibiting tumor progression. The drug is not cytotoxic like conventional chemotherapies. Instead it is cytostatic, and the NCCN guidelines recommend continuous administration of imatinib for metastatic GIST until there is evidence of progression [19]. The standard dose of imatinib is 400 mg, and it is able to be administered only once daily orally due to its circulating half-life of about 20 h.

In a large multicenter trial evaluating imatinib in the treatment of advanced and/or metastatic GISTs, tumors that were KIT-positive showed significant long-lasting response to the drug, with 85% surviving at 76 weeks after starting therapy [32]. However, in the same trial, 14% of the patients were found to have tumors resistant to imatinib, likely due to mutations in different proteins or in parts of the KIT protein structure that do not respond to imatinib binding. GIST patients with the most commonly observed KIT mutations in exon 11 have been found to respond to imatinib more favorably than those with KIT mutations involving exon 9 or wild-type GISTs [10]. This better response may be related to the fact that the binding site of imatinib is in close proximity to the portion of the KIT protein encoded by exon 11.

Dose escalation to 800 mg per day is the recommended first step in patients with unresectable or metastatic GIST that shows early evidence of progression on imatinib. GISTs with KIT mutations in exon 9 have also demonstrated a higher response rate and longer progression-free survival with this higher dose [10].

Preoperative imatinib therapy is recommended as a first-line therapy to reduce the size of large tumors. It is also recommended to reduce the size of tumors at sites where they may become adherent to adjacent tissue or organs and would initially require an extensive en bloc surgery with higher associated morbidity. For example, a duodenal GIST may initially require a pancreaticoduodenectomy, while a rectal GIST might require an abdominoperineal resection in which the rectal sphincter is unable to be spared. In cases like these, preoperative imatinib has been shown to lead to initially unresectable but localized GISTs becoming resectable and to less morbid organ- and function-preserving surgeries for initially marginally resectable tumors. Preoperative imatinib treatment has also been shown to reduce the risk of tumor rupture before or during surgery, which thus improves prognosis [33]. According to NCCN guidelines, preoperative imatinib should be monitored by imaging and continued until two consecutive restaging scans fail to demonstrate interval improvement in tumor burden [19].

Imatinib has been used as an adjuvant treatment following surgery. Even with complete surgical resection of GIST, it has high recurrence rate in the first 2 years after surgery. Imatinib was approved by the FDA in 2008 for treatment of GIST after complete surgical resection, as it was shown to significantly prolong survival in this setting. For patients who did not receive preoperative imatinib, imatinib following complete surgical resection is recommended for at least 36 months per the NCCN guidelines in patients who presented with higher-risk primary GISTs, namely, tumors larger than 5 cm with more than 5 mitoses per 50 HPF [19]. For patients who did receive preoperative imatinib, the NCCN guidelines recommend that imatinib therapy is considered following complete surgical resection if there had been an objective response to preoperative therapy. However, the duration of this treatment is uncertain, although there have been single- and multicenter trials that have shown benefit in continuing imatinib treatment for 2 years after surgery [19].

Sunitinib was the second receptor TKI approved by the FDA for the treatment of GIST in 2006. It is indicated in patients who have developed resistance to imatinib or have progression of disease following escalation of the daily imatinib dose from 400 to 800 mg. Sunitinib is a less specific inhibitory drug than imatinib and is structurally distinct. It inhibits KIT and PDGFR like imatinib, but also acts to inhibit vascular endothelial growth factor receptors 1–3 (VEGFR1–3), Fms-related tyrosine kinase 3, colony-stimulating factor 1R (CSF-1R), and the RET receptor [34]. Sunitinib thus possesses an antiangiogenic effect in addition to an antitumoral effect. Patients either receive a continuous daily oral 37.5 mg dose of sunitinib or 4 weeks of a continuous 50 mg dose followed by 2 weeks off treatment [35]. There is evidence suggesting that GISTs with KIT exon 9 mutations, wild-type GISTs, and pediatric GISTs, tumors which are typically less responsive to imatinib, are more sensitive to sunitinib [36]. Sunitinib has been considered as a first-line therapy for these particular tumors.

Regorafenib is an oral multikinase inhibitor that was approved by the FDA in 2013 for the treatment of imatinib- and sunitinib-resistant GIST. Regorafenib inihibits the activity of several protein kinases including KIT, RET, BRAF, and PDGFR and has an antiangiogenic effect with inhibition of VEGFR1-3 and TEK. Regorafenib was shown to increase progression-free survival in a phase 3 trial of GIST patients who had progressed on imatinib and sunitinib [37].

In patients with wild-type GISTs, which lack both KIT and PDGRA mutations, but who do exhibit mutations in the BRAF kinase, treatment with BRAF inhibitors has shown some success. Falchook et al. (2013) reported the first case study of treatment of one such patient using a BRAF inhibitor, dabrafenib [38]. Other novel drugs have been investigated in the treatment of metastatic GIST resistant to imatinib and sunitinib. Sorafenib is a TKI that exhibits inhibitory action against several kinases and has demonstrated significant disease control rates in several studies [39].

In addition to treatment with novel drugs, regimens of two complementary TKIs given daily in a rapidly alternating regimen are currently under investigation (e.g., NCT02164240 Phase Ib Study of SUnitinib Alternating With REgorafenib in Patients With Metastatic and/or Unresectable GIST (SURE) at clinicaltrials.gov) in patients with metastatic GIST resistant to imatinib. It is hoped that these regimens provide a broader spectrum of activity with better tolerated side effects than treatment with a single agent.

Assessing Response to Treatment

The standard imaging modality used to assess response to treatment in GIST is a contrast-enhanced CT scan. Tumor size is not a useful metric for CT evaluation of response of GIST to imatinib therapy, as it is for monitoring response to conventional cytocidal therapy used in most other tumors. Imatinib is a cytostatic therapy, and a significant decrease in size of GIST metastases may take months to develop, even if the tumor is responding to the drug. In fact, there may be a paradoxical increase in size of tumors responding to treatment due to development of intratumoral hemorrhage and/or cystic changes (Fig. 1). The common way of assessing tumor response to therapy has been change in unidimensional size measurements using Response Evaluation Criteria in Solid Tumors (RECIST) , where a partial response is defined as a 30% reduction in unidimensional measurements and progressive disease is defined as a 20% increase or the appearance of new lesions [40]. These definitions are obviously limited in the case of response of GIST to imatinib.
Fig. 1

(a) Baseline [18F]FDG-PET/CT of 67 year old man with GIST demonstrating the dominant [18F]FDG-avid hepatic metastasis. (b) PET/CT scan performed 2 months after start of imatinib demonstrates a slight increase in size of the dominant hepatic lesion, but a significant decrease in uptake and a decrease in associated density on CT scan. These findings are consistent with a significant metabolic response to therapy, despite the apparent increase in size of the dominant lesion

[18F]FDG-PET and [18F]FDG-PET/CT have proven valuable in assessing early response to both neoadjuvant and systemic treatment in GIST, especially in the time before changes in size are demonstrated on CT. In one of the earliest trials assessing response to imatinib in GIST, [18F]FDG-PET showed dramatic early decrease in [18F]FDG uptake, and thus metabolic activity, in tumors which preceded anatomic changes by weeks or months (Fig. 2) [41]. A significant change in metabolic response can be observed on [18F]FDG-PET in as little as hours after the start of imatinib therapy. This rapid response in metabolism was found to correspond at least in part with a decrease in glucose transporter 4 (GLUT4) expression in tumors on pathology prior to and following surgery in a study of patients receiving preoperative neoadjuvant imatinib [42].
Fig. 2

(a) Baseline [18F]FDG-PET/CT of 63 year old man with recently diagnosed metastatic gastrointestinal stromal tumor (GIST). There are multiple hepatic and peritoneal [18F]FDG-avid lesions. (b) PET/CT scan performed 1 week later after the start of treatment with imatinib. Marked reduction of FDG uptake is demonstrated in all lesions

The first phase 2 trial of imatinib [32] and many trials following it used criteria developed to assess metabolic response to treatment on [18F]FDG-PET scans developed by the European Organization for Research and Treatment of Cancer (EORTC) . Semiquantitative assessment of [18F]FDG uptake is made using the maximum standardized uptake value (SUVmax) of tumor sites. The EORTC guidelines defined partial response as a greater than 25% decrease in SUV of tumor on a scan compared to the pretreatment baseline scan, with progressive disease demonstrating a greater than 25% increase in SUV or new [18F]FDG-avid lesions [43]. The SUVmax is the most commonly used semiquantitative parameter for assessing response. When the EORTC guidelines for partial response to therapy have been met on early scans performed anywhere between 3 and 16 weeks after the start of imatinib therapy, several studies have demonstrated a correlation with more favorable outcome as defined by a longer time to treatment failure [44, 45]. Using a threshold of 40% reduction in baseline SUVmax on the first early PET scan has shown to be predictive of even longer time to treatment failure [46].

[18F]FDG-PET has shown advantages over diagnostic CT in certain cases in assessing for treatment response in metastatic GIST. Imprecise timing of the administered intravenous contrast may cause a liver lesion to be inconspicuous on the baseline scan, while an enhancing mesenteric lesion at baseline associated with bowel loops may also be similarly obscured. If these lesions respond to treatment on follow-up CT scans, their resulting decrease in attenuation may make them newly conspicuous and may lead to them being misinterpreted as new lesions. An [18F]FDG-PET/CT obtained at the same time as the follow-up diagnostic CT could help to properly characterize them as responding non-metabolically active lesions.

GIST has been found to have a characteristic pattern of response to imatinib treatment on CT scan. Heterogeneous high attenuation of tumor at baseline transitions to a more homogeneous pattern of lower attenuation (20–25 HU) during a response to treatment. These changes correlate histologically with a decrease in tumor vascularity and associated cystic and/or myxoid degeneration [47]. This change in attenuation usually occurs within 1–2 months [48]. Calcification and/or intratumoral hemorrhage can also occur, which leads to increased density and may mask response to therapy [49].

Alternative tumor response criteria on diagnostic CT proposed by Choi et al. (2007) incorporate change in tumor attenuation along with change in size and have been shown to have a similar correlation to metabolic response to imatinib as [18F]FDG uptake on [18F]FDG-PET. The Choi criteria define treatment response as a decrease in tumor attenuation on CT of 15% and/or a reduction of tumor unidimensional diameter of 10%. While both the Choi and RECIST criteria were found to have a 100% specificity in identifying patients demonstrating metabolic response on [18F]FDG-PET scan, the Choi criteria had a significantly higher sensitivity than the RECIST criteria, 97% versus 52% [50].

Dual-energy CT (DECT) allows for quantification of intratumoral intravenous iodinated contrast. DECT response criteria have been proposed and have been compared with the Choi CT criteria to determine whether they demonstrate better correlation with survival. DECT can quantify iodine-related attenuation (IRA), which can be considered a surrogate marker for perfusion and tumor vascularity. Meyer et al. compared DECT criteria to the Choi criteria in evaluating the response of GIST on tyrosine kinase inhibitor treatment [51]. The DECT criteria defined nonresponders as those with tumors that demonstrated an increase of 20% in both tumor size and IRA or a 50% increase in either tumor size or IRA, while responders were defined as all other patients. Those defined as responders by the DECT and Choi criteria demonstrated significantly longer progression-free survival than those defined as nonresponders. Of note, the DECT criteria, but not the Choi criteria, provided a statistically significant prediction of overall survival. One reason for this result was thought to be that intratumoral hemorrhage does not affect the quantification performed by DECT.

MRI has also been utilized in assessing the response of GIST to therapy. Some of the changes on MRI described in responding tumors are minimal to no contrast enhancement and T2 hyperintensity that may mimic that of cysts. Intratumoral hemorrhage can be more accurately differentiated from viable tumor on MRI than CT. Diffusion-weighted MRI images have also shown promise in characterizing early response to treatment. Diffusion-weighted MRI images demonstrate an increase ADC values compared to baseline values in tumors that are responding to treatment. Marked increase in the ADC values of lesions has been demonstrated on MRI just a week after the start of therapy and correlates with a favorable prognosis [52]. While early prognostic information provided by diffusion-weighted MRI may be on par with [18F]FDG-PET/CT in localized GIST involvement, [18F]FDG-PET/CT is better suited to evaluate the response of diffuse metastatic disease.

On all imaging modalities, the response of GIST to sunitinib and other TKIs is similar to the response seen on these modalities to imatinib. It should be kept in mind that sunitinib and certain other TKIs, unlike imatinib, have antiangiogenic effects on GIST which may result in intratumoral bleeding which may cause difficulty in CT interpretation. In these cases, [18F]FDG-PET/CT may be helpful to rule out the presence of any metabolically active tumor, while MRI would better characterize any blood products.

[18F]FDG-PET/CT continues to be utilized in trials involving novel TKIs due to its ability to identify early response to therapy and the positive prognostic significance of this early drop in metabolic activity.

Assessing Resistance to Treatment

Almost all patients treated with imatinib for metastatic or advanced GIST will inevitably develop resistance to the drug. This resistance can either develop early or late. Primary resistance is progression of disease within the first 6 months of imatinib treatment and has been found to occur in about 14% of patients [32]. There are several associations that have been described with primary resistance to imatinib including exon 9 KIT mutations, PDGFRA mutations, wild-type GISTs, and pediatric GISTs [53]. Response to initial imatinib therapy has been demonstrated consistently earlier on [18F]FDG-PET/CT than on diagnostic CT, with decreases in metabolic activity often evident within the first week of treatment, while CT changes suggestive of response, including decrease in attenuation, may take up to a month or more to clearly develop. Therefore, [18F]FDG-PET/CT may be of particular value in determining if a GIST has primary resistance to imatinib compared to other modalities. Lack of significant decrease in [18F]FDG uptake from baseline of tumor on a [18F]FDG-PET/CT scan performed during the first month of therapy is highly predictive of primary resistance, and this result should prompt clinicians to escalate the dose of imatinib [54]. If there is continued resistance demonstrated on scans after dose escalation, treatment with the second-line TKI sunitinib would be considered (Fig. 3).
Fig. 3

(a) 44 year old man with metastatic GIST being treated with imatinib. [18F]FDG uptake is demonstrated on PET scan in hepatic and mesenteric lesions (red arrows). Black arrows represent sites of colonic uptake unrelated to GIST without CT correlates. (b) Follow up PET scan after 6 weeks demonstrates an increase in size and uptake of the hepatic and mesenteric lesions, as well as two newly [18F]FDG-avid hepatic metastases (red arrow), findings consistent with progression of disease. It was decided clinically to escalate the dose of imatinib. (c) Follow up PET scan after 3 weeks demonstrates a decrease in [18F]FDG uptake in hepatic lesions, but an increase in uptake of the mesenteric lesion (red arrow). Although the hepatic disease did show response to the higher dose of imatinib, the mesenteric lesion continued to progress, indicating that this lesion had developed resistance to imatinib. (Of note, the focal sites of colonic uptake unrelated to GIST without CT correlates persist, and are highly suspicious for underlying adenomatous polyps)

Secondary resistance to imatinib occurs in patients who initially demonstrated a good response to treatment for at least 6 months. Secondary resistance to imatinib usually develops after a median of 2 years [55]. Usually secondary KIT mutations are responsible, often in exons [13, 14, 17] that code for portions of the intracytoplasmic portions of the KIT protein [36].

Findings of secondary resistance described on CT include new metastatic lesions and increase of size and/or attenuation of preexisting lesions. A characteristic pattern of tumor recurrence on CT or MRI is the so-called new “nodule-within-a-mass” appearance [56]. Here, recurrent tumor is defined as a new enhancing soft-tissue nodule or nodules on CT or MRI or within a cystic mass.

[18F]FDG-PET/CT can identify secondary resistance by the reappearance of [18F]FDG uptake within a previous lesion which was not [18F]FDG-avid on treatment. In patients with a large number of metastases, only a few lesions may demonstrate new [18F]FDG uptake, yet this appearance would still be consistent with development of resistance. [18F]FDG-PET/CT has the ability to demonstrate that resistance does not just arise throughout the entirety of discrete lesions. A small new focus of uptake within an otherwise photopenic lesion can suggest resistance developing within a subpopulation of cells within a single lesion (Fig. 4). These findings on [18F]FDG-PET scans suggest that patients who develop secondary resistance have two biologically distinct populations of tumor, one that is still responding to imatinib and one which has developed mutations granting them resistance to the drug. This phenomenon was illustrated during a phase I/II clinical trial for sunitinib, where the trial protocol required patients to be off imatinib for a period of time (up to 3 weeks) before starting sunitinib. New baseline [18F]FDG-PET scans were obtained with the patients taking neither drug. In addition to an increase in [18F]FDG uptake seen in the lesions already determined to have imatinib-resistant GIST, a “flare” of new uptake was observed in many of the other previously non-[18F]FDG-avid lesions (Fig. 5) [54]. Discontinuation of imatinib had lead to reactivation of glycolytic metabolism in the population of tumor cells that continued to respond to the drug.
Fig. 4

(a) Same patient as in Fig. 3, undergoing treatment with escalated dose of imatinib. (b) Three week follow up PET scan not only demonstrates new [18F]FDG uptake in another mesenteric lesion (red arrow) suggesting resistance to imatinib, but there is new increased [18F]FDG uptake in the anterior aspect of the dominant hepatic lesion (yellow arrow), consistent with resistance to imatinib within a subpopulation of cells within an otherwise treated lesion. The clinical decision was made at this point to discontinue imatinib in anticipation of starting treatment with sunitinib

Fig. 5

(a) 48 year old man with metastatic GIST with development of a hepatic lesion resistant to imatinib (black arrow). (b) PET scan performed 3 weeks later prior to the start of sunitinib. Not only has the resistant lesion within the liver increased in [18F]FDG uptake (black arrow), but there has been reactivation of metabolic activity in hepatic and mesenteric lesions which were previously responding to imatinib (red arrows)

Surveillance

GIST has a recurrence rate as high as 90% in the first 2 years, even after a successful complete surgical resection. GIST typically recurs in the liver or peritoneum. The current NCCN guidelines recommend contrast-enhanced CT for surveillance of GIST, especially in those initially characterized as high risk [19]. Patients are typically scanned in 3- to 6-month intervals. [18F]FDG-PET/CT is usually only utilized when there is an ambiguous finding on CT or when treatment with a new TKI is considered and a baseline [18F]FDG-PET scan is warranted.

In addition to detecting recurrent or progressive disease, surveillance scans can also identify side effects of TKI therapy. One of the more common side effects of imatinib seen on imaging is the third spacing of fluid within the body which can manifest as pleural effusions, pulmonary edema, ascites, and/or anasarca. As development of ascites would normally raise suspicion for the presence of new peritoneal disease, care must be taken in determining whether this is just a fluid overload-related side effect of the treatment. Other side effects of the various TKIs that may manifest on imaging include hepatic steatosis, hepatitis, pancreatitis, ischemic colitis, pneumatosis, and bowel perforation [57, 58].

References

  1. 1.
    Miettinen M, Lasota J. Gastrointestinal stromal tumors: definition, clinical, histological, immunohistological, and molecular genetic features and differential diagnosis. Virchow’s Arch. 2001;438:1–12.CrossRefGoogle Scholar
  2. 2.
    Miettinen M, et al. Evaluation of malignancy and prognosis of gastrointestinal stromal tumors: a review. Hum Pathol. 2002;33:478–83.CrossRefPubMedGoogle Scholar
  3. 3.
    Ma GL, et al. Epidemiology of gastrointestinal stromal tumors in the era of histology codes: results of a population-based study. Cancer Epidemiol Biomarkers Prev. 2015;24:298.CrossRefPubMedGoogle Scholar
  4. 4.
    Kindblom LG, Remotti HE, Aldenborg F, Meis-Kindblom JM. Gastrointestinal pacemaker cell tumor (GIPACT): gastrointestinal stromal tumors show phenotypic characteristics of the interstitial cells of Cajal. Am J Pathol. 1998;152:1259–69.PubMedPubMedCentralGoogle Scholar
  5. 5.
    Miettinen M, Lasota J. Gastrointestinal stromal tumors: pathology and prognosis at different sites. Semin Diagn Pathol. 2006;23:70–83.CrossRefPubMedGoogle Scholar
  6. 6.
    Tran T, Davila JA, El-Serag HB. The epidemiology of malignant gastrointestinal stromal tumors: an analysis of 1,458 cases from 1992 to 2000. Am J Gastroenterol. 2005;100:162.CrossRefPubMedGoogle Scholar
  7. 7.
    Hirota S, Isozaki K, Moriyama Y, et al. Gain-of-function mutations of c-kit in human gastrointestinal stromal tumors. Science. 1998;279:577–80.CrossRefPubMedGoogle Scholar
  8. 8.
    Heinrich MC, Corless CL, Duensing A, et al. PDGFRA activating mutations in gastrointestinal stromal tumors. Science. 2003;299:708–10.CrossRefPubMedGoogle Scholar
  9. 9.
    Miettinen M, Lasota J. Gastrointestinal stromal tumors: review on morphology, molecular pathology, prognosis, and differential diagnosis. Arch Pathol Lab Med. 2006;130:1466–78.PubMedGoogle Scholar
  10. 10.
    Marrari A, Wagner AJ, Hornick JL. Predictors of response to targeted therapies for gastrointestinal stromal tumors. Arch Pathol Lab Med. 2012;136:483–9.CrossRefPubMedGoogle Scholar
  11. 11.
    Corless CL, et al. PDGFRA mutations in gastrointestinal stromal tumors: frequency, spectrum, and in vitro sensitivity to imatinib. J Clin Oncol. 2005;23:5357.CrossRefPubMedGoogle Scholar
  12. 12.
    Miettinen M, et al. Gastrointestinal stromal tumors in patients with neurofibromatosis 1: a clinicopathologic and molecular genetic study of 45 cases. Am J Surg Pathol. 2006;30:90.CrossRefPubMedGoogle Scholar
  13. 13.
    Pappo AS, Janeway KA. Pediatric gastrointestinal stromal tumors. Hematol Oncol North Am. 2009;23:15.CrossRefGoogle Scholar
  14. 14.
    Bachet JB, et al. Diagnosis, prognosis and treatment of patients with gastrointestinal stromal tumour (GIST) and germline mutation of KIT exon 13. Eur J Cancer. 2013;49:2531.CrossRefPubMedGoogle Scholar
  15. 15.
    Agaram NP, et al. Novel V600E BRAF mutations in imatinib-naive and imatinib-resistant gastrointestinal stromal tumors. Genes Chromosomes Cancer. 2008;47:886–96.CrossRefGoogle Scholar
  16. 16.
    Corless CL, Fletcher JA, Heinrich MC. Biology of gastrointestinal stromal tumors. J Clin Oncol. 2004;22(18):3813–25.CrossRefPubMedGoogle Scholar
  17. 17.
    Miettinen M, Sobin LH, Lasota J. Gastrointestinal stromal tumors of the stomach: a clinicopathologic, immunohistochemical, and molecular genetic study of 1765 cases with long-term follow-up. Am J Surg Pathol. 2005;29:52.CrossRefPubMedGoogle Scholar
  18. 18.
    Miettinen M, Makhlouf H, Sobin LH, Lasota J. Gastrointestinal stromal tumors of the jejunum and ileum: a clinicopathologic, immunohistochemical, and molecular genetic study of 906 cases before imatinib with long-term follow-up. Am J Surg Pathol. 2006;30:477.CrossRefPubMedGoogle Scholar
  19. 19.
    von Mehren M. et al. NCCN clinical practice guidelines in oncology. Soft tissue sarcoma. Version 2.2016. National Comprehensive Cancer Network website. http://www.nccn.org/professionals/physician_gls/f_guidelines.asp.
  20. 20.
    Singer S, et al. Prognostic value of KIT mutation type, mitotic activity, and histologic subtype in gastrointestinal stromal tumors. J Clin Oncol. 2002;20:3898.CrossRefPubMedGoogle Scholar
  21. 21.
    Martin J, et al. Deletions affecting codons 557–558 of the c-KIT gene indicate a poor prognosis in patients with completely resected gastrointestinal stromal tumors: a study by the Spanish group for sarcoma research (GEIS). J Clin Oncol. 2005;23:6190.CrossRefPubMedGoogle Scholar
  22. 22.
    Antonescu CR, et al. Association of KIT exon 9 mutations with nongastric primary site and aggressive behavior: KIT mutation analysis and clinical correlates of 120 gastrointestinal stromal tumors. Clin Cancer Res. 2003;9(9):3329.PubMedGoogle Scholar
  23. 23.
    Yu MH, et al. MRI features of gastrointestinal stromal tumors. AJR Am J Roentgenol. 2014;203:980–91.CrossRefPubMedGoogle Scholar
  24. 24.
    Burkill GJ, et al. Malignant gastrointestinal stromal tumor: distribution, imaging features, and pattern of metastatic spread. Radiology. 2003;226:527–32.CrossRefPubMedGoogle Scholar
  25. 25.
    Kim HC, et al. Small gastrointestinal stromal tumors with focal areas of low attenuation on CT: pathologic correlation. Clin Radiol. 2005;60:384–8.CrossRefPubMedGoogle Scholar
  26. 26.
    Sepe PS, Brugge WR. A guide for the diagnosis and management of gastrointestinal stromal tumors. Nat Rev Gastroenterol Hepatol. 2009;6:363–71.CrossRefPubMedGoogle Scholar
  27. 27.
    Gayed I, et al. The role of F-18 FDG PET in staging and early prediction of response to therapy of recurrent gastrointestinal stromal tumors. J Nucl Med. 2004;45:17–21.PubMedGoogle Scholar
  28. 28.
    Blodgett TM, et al. PET/CT: form and function. Radiology. 2007;242:360–85.CrossRefPubMedGoogle Scholar
  29. 29.
    Edmonson JH, et al. Contrast of response to dacarbazine, mitomycin, doxorubicin, and cisplatin (DMAP) plus GM-CSF between patients with advanced malignant gastrointestinal stromal tumors and patients with other advanced leiomyosarcomas. Cancer Invest. 2002;20:605–12.CrossRefPubMedGoogle Scholar
  30. 30.
    Trent JC, et al. A two-arm phase II study of temozolomide in patients with advanced gastrointestinal stromal tumors and other soft tissue sarcomas. Cancer. 2003;98:2693–9.CrossRefPubMedGoogle Scholar
  31. 31.
    Pierie JEN, et al. The effect of surgery and grade on outcome of gastrointestinal stromal tumors. Arch Surg. 2001;136:383–9.CrossRefPubMedGoogle Scholar
  32. 32.
    Demetri GD, et al. Efficacy and safety of imatinib mesylate in advanced gastrointestinal stromal tumors. N Engl J Med. 2002;347:472–80.CrossRefPubMedGoogle Scholar
  33. 33.
    Eisenberg BL, et al. Phase II trial of neoadjuvant/adjuvant imatinib mesylate (IM) for advanced primary and metastatic/recurrent operable gastrointestinal stromal tumor (GIST): early results of RTOG 0132/ACRIN 6665. J Surg Oncol. 2009;99:42–7.CrossRefPubMedPubMedCentralGoogle Scholar
  34. 34.
    Demitri GD, et al. Efficacy and safety of sunitinib in patients with advanced gastrointestinal stromal tumor after failure of imatinib: a randomised controlled trial. Lancet. 2006;368:1329–38.CrossRefGoogle Scholar
  35. 35.
    George S, et al. Clinical evaluation of continuous daily dosing of sunitinib malate in patients with advanced gastrointestinal stromal tumour after imatinib failure. Eur J Cancer. 2009;45:1959–68.CrossRefPubMedGoogle Scholar
  36. 36.
    Heinrich MC, et al. Molecular correlates of imatinib resistance in gastrointestinal stromal tumors. J Clin Oncol. 2006;24:4764–74.CrossRefPubMedGoogle Scholar
  37. 37.
    Demitri GD, et al. Efficacy and safety of regorafenib for advanced gastrointestinal stromal tumours after failure of imatinib and sunitinib (GRID): and international, multicentre, randomised, placebo-controlled, phase 3 trial. Lancet. 2013;381:295–302.CrossRefGoogle Scholar
  38. 38.
    Falchook, et al. BRAF mutant gastrointestinal stromal tumor: first report of regression with BRAF inhibitor dabrafenib (GSK2118436) and whole exomic sequencing for analysis of acquired resistance. Oncotarget. 2013;4:310–5.CrossRefPubMedPubMedCentralGoogle Scholar
  39. 39.
    Park SH, et al. Sorafenib in patients with metastatic gastrointestinal stromal tumors who failed two or more prior tyrosine kinase inhibitors: a phase II study of Korean gastrointestinal stromal tumors study group. Invest New Drugs. 2012;30:2377–83.CrossRefPubMedGoogle Scholar
  40. 40.
    Therasse P, et al. New guidelines to evaluate the response to treatment in solid tumors. European organization for research and treatment of cancer, national cancer institute of the United States, national cancer institute of Canada. J Natl Cancer Inst. 2000;92:205–16.CrossRefGoogle Scholar
  41. 41.
    Van den Abbeele AD, et al. GIST Collaborative PET Study Group. F18-FDG-PET provides early evidence of biological response to STI571 in patients with malignant gastrointestinal stromal tumors (GIST). Proc Am Soc Clin Oncol. 2001;20:362a.Google Scholar
  42. 42.
    Van den Abbeele AD, et al. ACRIN 6665/RTOG 0132 phase II trial of neoadjuvant imatinib mesylate for operable malignant gastrointestinal stromal tumor: monitoring with F-18-FDG PET and correlation with genotype and GLUT4 expression. J Nucl Med. 2011;53:567–74.CrossRefGoogle Scholar
  43. 43.
    Young H, et al. Measurement of clinical and subclinical tumour response using [18F]-fluorodeoxyglucose and positron emission tomography: review and 1999 EORTC recommendations. European organization for research and treatment of cancer (EORTC) PET study group. Eur J Cancer. 1999;35:1773–82.CrossRefPubMedGoogle Scholar
  44. 44.
    Gayed I, et al. The role of 18F-FDG PET in staging and early prediction of response to therapy of recurrent gastrointestinal stromal tumors. J Nucl Med. 2004;45:17–21.PubMedGoogle Scholar
  45. 45.
    Goerres GW, et al. The value of PET, CT, and in-line PET/CT in patients with gastrointestinal stromal tumors: long term outcome of treatment with imatinib mesylate. Eur J Nuc Med Mol Imaging. 2005;32:153–62.CrossRefGoogle Scholar
  46. 46.
    Holdsworth CH, et al. CT and PET: Early prognostic indicators of response to imatinib mesylate in patients with gastrointestinal stromal tumor. AJR Am J Roentgenol. 2007;189:W324–30.CrossRefPubMedGoogle Scholar
  47. 47.
    Abdulkader I, et al. Pathological changes related to imatinib treatment in a patient with a metastatic gastrointestinal stromal tumour. Histopathology. 2005;46:470–2.CrossRefPubMedGoogle Scholar
  48. 48.
    Hong X, et al. Gastrointestinal stromal tumor: role of CT in diagnosis and in response evaluation and surveillance after treatment with imatinib. Radiographics. 2006;26:481–95.CrossRefPubMedGoogle Scholar
  49. 49.
    Tirumani SH, et al. Imatinib and beyond in gastrointestinal stromal tumors: a radiologist’s perspective. AJR Am J Roentgenol. 2013;201:801–9.CrossRefPubMedGoogle Scholar
  50. 50.
    Choi H, et al. Correlation of computed tomography and positron emission tomography in patients with metastatic gastrointestinal stromal tumor treated at a single institution with imatinib mesylate: proposal of new computed tomography response criteria. J Clin Oncol. 2007;25:1753–9.CrossRefPubMedGoogle Scholar
  51. 51.
    Meyer M, et al. CT-based response assessment of advanced gastrointestinal stromal tumor: dual energy CT provides a more predictive imaging biomarker of clinical benefit than RECIST or Choi criteria. Eur J Radiol. 2013;82(6):923–8.CrossRefPubMedGoogle Scholar
  52. 52.
    Tang L, et al. Gastrointestinal stromal tumors treated with imatinib mesylate: apparent diffusion coefficient in the evaluation of therapy response in patients. Radiology. 2011;258:729–38.CrossRefPubMedGoogle Scholar
  53. 53.
    Benjamin RS, et al. Gastrointestinal stromal tumors II: medical oncology and tumor response assessment. Semin Oncol. 2009;36:302–11.CrossRefPubMedGoogle Scholar
  54. 54.
    Van den Abbeele AD. The lessons of GIST—PET and PET/CT: a new paradigm for imaging. Oncologist. 2008;13:8–13.CrossRefPubMedGoogle Scholar
  55. 55.
    Verwejj J, et al. Progression-free survival in gastrointestinal stromal tumors with high-dose imatinib: randomised trial. Lancet. 2004;354(9440):1127–34.CrossRefGoogle Scholar
  56. 56.
    Shankar S, et al. Gastrointestinal stromal tumor: new nodule-within-a-mass pattern of recurrence after partial response to imatinib mesylate. Radiology. 2005;235:892–8.CrossRefPubMedGoogle Scholar
  57. 57.
    Howard, et al. Decade of molecular targeted therapy: abdominal manifestations of drug toxicities—what radiologists should know. AJR Am J Roentgenol. 2012;199:58–64.CrossRefPubMedGoogle Scholar
  58. 58.
    Shingare, et al. Pneumatosis intestinalis and bowel perforation associated with molecular targeted therapy: an emerging problem and the role of radiologists in its management. AJR Am J Roentgenol. 2012;199:1259–65.CrossRefGoogle Scholar

Copyright information

© Springer International Publishing AG 2016

Authors and Affiliations

  • Christopher G. Sakellis
    • 1
    • 2
    • 3
    • 4
    • 5
  • Heather A. Jacene
    • 1
    • 2
    • 3
    • 4
    • 5
  • Annick D. Van den Abbeele
    • 1
    • 2
    • 3
    • 4
    • 5
    Email author
  1. 1.Department of ImagingDana-Farber Cancer InstituteBostonUSA
  2. 2.Department of RadiologyBrigham and Women’s HospitalBostonUSA
  3. 3.Harvard Medical SchoolBostonUSA
  4. 4.Center for Biomedical Imaging in OncologyDana-Farber Cancer InstituteBostonUSA
  5. 5.Tumor Imaging Metrics CoreDana-Farber/Harvard Cancer CenterBostonUSA

Personalised recommendations