Metastatic well or moderately differentiated neuroendocrine tumors of the gastrointestinal tract and lung (NETs) are a fascinating and markedly heterogeneous group of generally indolent, but relentless cancers. These unusual (but not rare) malignancies can be managed reasonably with a range of therapeutic options from observation to systemic therapies that are generally well tolerated but only modestly active—from minimally invasive regional therapies generally focused on hepatic dominant metastases to aggressive surgical palliation with multilobe hepatic resection or even liver transplantation.
It is humbling to realize that the management of NETs has not progressed far from the excellent pearls of clinical behavior and management outlined by Charles Moertel from the Mayo Clinic in the inaugural volume of the Journal of Clinical Oncology 25 years ago, and in Dr Moertel’s Karnofsky memorial lecture “An Odyssey in the Land of Small Tumors” in 1987. This article should be required reading for anyone providing care to these patients .
This excellent review by Phan and Yao outlines the remarkable clinical research activity of the past several years, with the identification of several potentially promising targeted agents after many years of minimal progress. As more opportunities arise to identify new therapies for NETs, the heterogeneity of these malignancies and the importance of good basic clinical trial design based on biology, prognostic factors, and standardized endpoints will help with our interpretation of what therapies are “promising.”
Spectrum of Diseases
The many patients who are diagnosed with and are treated for “carcinoid,” who are referred to tertiary centers, and who are enrolled in clinical trials, may actually have several different diseases in the NET spectrum. The most widely appreciated patient scenario is that of the metastatic small intestinal carcinoid, which frequently causes some degree of flushing, diarrhea, palpitations, and hypertension (the carcinoid syndrome), and the potential for right-sided heart valve sclerosis and incompetence (carcinoid heart disease). The growth of this disease itself is generally very indolent.
NETs vary with regard to natural history, responsiveness to standard chemotherapy, and biology, with potentially different susceptibilities to novel agents. By standard light microscopy and immunohistochemical staining of a liver biopsy, a small intestinal carcinoid, pancreatic islet cell carcinoma (ICC), and rectal carcinoid may all be called “carcinoid.” However, it is important to differentiate between these entities clinically (eg, rectal carcinoids do not cause carcinoid syndrome and may carry a more aggressive natural history than small intestinal carcinoid; ICC may be much more sensitive to systemic chemotherapy) and, increasingly, account for these different entities in clinical trials to allow optimal interpretation of results. This review article provides a nice overview of what is known of the different biology of various NETs, but also demonstrates there is much to learn.
Promising New Agents
As outlined by the authors, the most promising new targeted agents to date include inhibitors of mammalian target of rapamycin (mTOR) and angiogenesis. Multicenter phase II trials in patients with carcinoid and ICC treated with sunitinib (Sutent) and sorafenib(Drug information on sorafenib) (Nexavar), which likely work through predominantly antiangiogenic pathways, yielded similar results, ie, a 2% to 17% partial response rate by Response Evaluation Criteria in Solid Tumors (RECIST). Bevacizumab(Drug information on bevacizumab) (Avastin) has also demonstrated a modest response rate in carcinoid tumors when given along with long-acting octreotide(Drug information on octreotide) (Sandostatin LAR) in a small study.
Everolimus (RAD-001), an oral mTOR inhibitor, was evaluated in a single-center phase II trial, given in combination with octreotide LAR. This trial demonstrated remarkable results, including a partial response rate of 20% overall (17% in carcinoid, 27% in ICC) and median progression-free survival (PFS) of 12 to 14 months. However, a number of these patients had not received prior octreotide, there was no requirement for progressive disease on study, and all were able to travel to a single center for therapy. In contrast, the mTOR inhibitor temsirolimus (Torisel) was evaluated in a multicenter phase II trial that required documented disease progression within 6 months prior to study therapy and noted a response rate of only 5.6% and median time to progression of 6 months. With these contrasting results, it would be difficult to know whether mTOR inhibition is “promising” or not. Do these agents require concomitant octreotide for best results? Is one drug superior to the other?
In this context, a large multicenter phase II trial of everolimus in metastatic ICC in patients with prior chemotherapy exposure and progressive disease resulted in response rates of 4% with concomitant octreotide (for patients already on this agent), and 8% as a single agent, similar to the temsirolimus results. Median PFS was 13 months with octreotide, but only 9 months with everolimus as a single agent. This interesting trial suggests that the evaluation of any new therapy for NET needs to be interpreted in light of which patient population is being treated. Also, does the apparent improved outcome with combined octreotide and everolimus suggest synergy of these agents, perhaps through octreotide suppression of signaling pathways upstream of mTOR? Or does its use suggest that these patients have a functioning ICC with symptomatic hormone production, which may have a more indolent natural history, and/or be found at a time of less disease bulk than nonfunctioning tumors, and may need to be stratified for in future trials?
It should be noted that the radiographic evaluation of liver metastases in NETs can be very difficult. Given the intense vascularity of the metastatic lesions, standard contrast-enhanced computed tomography (CT) scans during the standard portal venous phase of the study may greatly underestimate or obscure lesions. A biphasic contrast-enhanced CT scan with arterial phase images or magnetic resonance imaging (MRI) scan results in better imaging that is more reproducible for comparison through time, and should be mandated in any clinical trials conducted in NET.
High rates of stable disease were also reported across these studies. Given the indolent natural history of these tumors, even many patients observed off therapy would not meet RECIST criteria for progression at 2- to 3-month reevaluation, and thus stable disease is uninformative, unless quantified as “prolonged,” perhaps at 6 or 12 months.
In an attempt to gain further insight into activity in these studies, data from the sunitinib and initial everolimus studies were displayed as waterfall plots, which appear to show that many patients had some small reduction in tumor size with therapy as “best response.” However, this would require just one measurement, without confirmation on subsequent scan, that registers a difference in lesion size as small as a few millimeters. The presentation of the sorafenib trial provided data regarding confirmed minor responses (20% to 29% reduction in the aggregated diameters of measurable lesions) of 7% for carcinoid and 14% for ICC. It is unclear whether either method yields valuable information regarding clinical activity.
A major challenge moving forward will be the establishment of valid clinical endpoints, as overall survival is too prolonged to be useful in most NET studies. Partial response may also not be a useful measure of prolonged disease control, which may be a more important contribution of new therapies.
In the phase II trial of gefitinib(Drug information on gefitinib) (Iressa), few objective responses were seen, but disease progression by RECIST was required for entry into the study. Rapid disease progression was seen in many patients, emphasizing the poor-prognosis group on this study, and PFS at 6 months—the primary endpoint to this study—was only 51% in carcinoid patients and 28% in ICC. However, the interval between scans documenting pretreatment progression was known for all patients, as this was an eligibility criterion. We could then use this baseline rate of progression and look at time to progression while on therapy. Time to progression on study exceeded the prestudy interval by over 4 months in 32% and 17% of carcinoid and ICC patients, respectively, despite no objective responses. This approach might suggest meaningful activity in select patients .
With the current and future availability of promising agents for NET, we need to start investigating the potential for rational combinations of these agents as well as the development of new agents. Ideally, investigators can come together and agree on a framework for future clinical trials. Description of the population in terms of primary site of disease, functioning tumors, prior therapies, progressive disease by RECIST at enrollment, pathologic grade or Ki-67 level, and use of octreotide, among other factors, should be defined. Carcinoid tumors and ICC should not be treated as one disease. Clinical endpoints of response, prolonged stability (perhaps at 6 to 12 months, if prior progression), and symptom improvement should be reported in a consistent fashion.
It may be that the use of randomized phase II designs will be most helpful to screen combination treatments or new agents with a contemporary control and plan for larger studies. This type of discussion is going on in the neuroendocrine task force of the National Cancer Institute’s GI Intergroup steering committee. Participation in clinical trials to improve therapy for all of our NET patients should be a priority.
Financial Disclosure: The authors have no significant financial interest or other relationship with the manufacturers of any products or providers of any service mentioned in this article.