Meta‐Analysis of Randomized Clinical Trials Comparing Active Treatment with Placebo in Metastatic Neuroendocrine Tumors

Neuroendocrine tumors (NETs) are slow‐growing and often asymptomatic and the optimal upfront management approach is controversial. Despite the lack of evidence‐based data, the watch‐and‐wait strategy is still included in clinical guidelines. The present article reports the results of a meta‐analysis of phase III clinical trials that have compared active treatment to placebo in patients with metastatic NETs.

Abstract

Background.

Most guidelines still recommend active surveillance for patients with asymptomatic, unresectable neuroendocrine tumors (NETs). However, recent findings from several randomized placebo‐controlled trials suggest that most patients would benefit from active treatment. We conducted a meta‐analysis of pooled outcomes from clinical trials in which an active treatment arm was compared with placebo to determine whether active treatment provides a survival advantage.

Materials and Methods.

This meta‐analysis evaluated six trials that compared a medication with placebo in patients with an asymptomatic, metastatic NET. The trials were heterogenous with regard to the active medication (octreotide, lanreotide, sunitinib, everolimus, Lu‐Dotatate) and tumor localizations (gastrointestinal, pancreas, lung). Overall survival (OS) and progression‐free survival (PFS) for the placebo and active treatment arms were obtained from individual trial data and combined to obtain pooled outcomes.

Results.

The individual trials all reported significantly better PFS outcomes for active treatment. The pooled data confirmed this advantage. At months 3, 6, 12, 18, and 24, pooled PFS rates for the placebo and treatment arms, respectively, were 92.9% versus 96.9%; 54.3% versus 83.7%; 35.5% versus 68.5%; 25.1% versus 54.7%; and 17.7% versus 61.0%. OS was also higher in the active treatment groups. At months 6, 12, 24, 36, 48, and 60, OS rates (placebo vs. active treatment), respectively, were 88.1% versus 93.4%; 84.1% versus 86.2%; 67.4% versus 76%; 56.6% versus 64.4%; 49.9% versus 61.0%; and 41.7% versus 45.9%.

Conclusion.

This meta‐analysis confirms findings from recent clinical trials indicating that active treatment yields better survival outcomes than placebo. Importantly, these findings were obtained across a wide range of patient profiles and diverse medical treatments for metastatic NETs. Given the lack of reliable prognostic factors to determine a priori which patients are unlikely to benefit from active treatment, these findings support early treatment in most patients.

Implications for Practice.

Although most guidelines still recommend active surveillance for patients diagnosed with metastatic neuroendocrine tumors, the results of this meta‐analysis, together with recent data from key clinical trials, suggest that most patients could benefit from upfront active treatment. However, more data are needed to confirm this.

Introduction

Neuroendocrine tumors (NETs) are relatively rare neoplasms that can occur in a wide range of sites, including the gastrointestinal (GI) tract, pancreas, and lung [1]. Although the clinical behavior of these tumors is often heterogenous and unpredictable, in most cases these tumors are slow‐growing and often asymptomatic. For this reason, most NETs are locally advanced or metastatic at diagnosis and therefore unsuitable for radical‐intent surgery [2].

The optimal medical management of asymptomatic, unresectable NETs is controversial [3], in part because a substantial proportion of patients can remain progression free—even without treatment—for months or even years. For this reason, many authors and clinical guidelines [1], [4], [5] recommend active surveillance in asymptomatic patients with low‐grade, nonfunctional tumors. However, this approach has an important drawback: the unpredictable clinical course of these tumors, which tend to progress, even when the biological characteristics of the tumor are favorable [6]. Unfortunately, there is no clear clinical parameter or biomarker to predict when (or if) the tumor growth rate is likely to change. Although active surveillance is widely used in these cases, solid evidence to support this approach is lacking [6], [7]. Moreover, the recent findings from the CLARINET [8] and PROMID [9], [10] randomized controlled trials (RCTs) demonstrating an antiproliferative effect for somatostatin analogues (SSAs)—the mainstay of treatment for NETs—provide further support for the use of SSAs even in cases with favorable prognostic features [3], [6], [11]. Although the results of these two trials have increased support for the early administration of SSAs, the optimal management of these patients remains controversial [6] given that patients initially randomized to placebo in the CLARINET trial remained free of progression for an extended period (median, 18 months).

In this context, we performed a meta‐analysis of pooled outcomes from all of the published RCTs that have compared an active treatment with placebo in patients with metastatic NETs. The main aim was to assess progression‐free survival (PFS) and overall survival (OS) in the placebo arms to determine if the reported advantage of active treatment over placebo was maintained in pooled data across a wide range of patient profiles and medical treatments.

Materials and Methods

First, we identified all RCTs that compared medical treatment with placebo in metastatic NETs, regardless of the tumor site. To be eligible for inclusion in the present meta‐analysis, these studies were required to report PFS and/or OS data and include survival curves.

Six trials met the aforementioned inclusion criteria for this meta‐analysis: (a) PROMID (octreotide) [9], [10]; (b) CLARINET (lanreotide) [8]; (c) SU‐1111 (sunitinib malate) [12], [13]; (d) RADIANT‐3 (everolimus) [14]; (e) RADIANT‐4 (everolimus) [15]; and (f) NETTER‐1 (177 Lu‐Dotatate) [16]. Follow‐up data for the PROMID and the SU‐1111 trials were reported, respectively, by Rinke et al. [10] and Faivre et al. [13]. Updated results for the NETTER‐1 study were reported by Strosberg et al. [17] in 2018.

These trials compared a wide range of drugs in patients with diverse tumor localizations (GI, pancreas, and lung). In the included trials, PFS was defined as time to disease progression according to the RECIST criteria. Table 1 summarizes these trials.

Table 1. Randomized controlled trials comparing active treatment with placebo for the treatment of metastatic neuroendocrine tumors.

^aNot a true placebo arm because these patients received octreotide LAR.

Statistical Analysis

The Kaplan‐Meier curves for OS and PFS for placebo and treated groups were obtained. From the survival curves, we used the web plot digitizer program (available at http://arohatgi.info/WebPlotDigitizer/) to estimate the percentage of surviving patients at the following time points: months 6, 12, 24, 36, 48, and 60 for OS and months 3, 6, 12, 18, and 24 for PFS. A random effects meta‐analysis was carried out at each of the observation points. Because survival curves were not available from all of the studies at all time points, the analysis for each time point (e.g., month 6, 12, etc.) included only the studies that provided data for that time point.

Next, we combined the percentage of survivors in each group (placebo and treatment) at months 6, 12, 18, and 24 according to published recommendations [18]. Using these data, we created tables with survival estimates and 95% confidence intervals. Studies that included more cases were weighted more heavily. The DerSimonian‐Laird random effects model [19], which accounts for both intra‐ and interstudy variability, was used to pool the data and calculate these estimates. All analyses were performed with the SPSS statistical software program (IBM‐SPSS Inc., New York, NY) and the Stata software program, version 13.1 (StataCorp, College Station, TX).

Results

The six trials in this meta‐analysis included a total of 1,401 patients; of these, 747 received active treatment and 644 placebo.

Progression‐Free Survival: Placebo Groups

PFS in the placebo groups was based on the following five studies (Table 2): PROMID [9], CLARINET [8], RADIANT‐3 [20], RADIANT‐4 [15], and SU‐1111 [13].

Table 2. Estimated pooled PFS in placebo groups at months 3, 6, 12, 18, and 24.

Data are presented as estimated PFS, % (95% CI).

Abbreviations: CI, confidence interval; N/A, data not available/not applicable; PFS, progression‐free survival.

Overall Survival: Placebo Groups

The following four studies (Table 3) provided data on OS in the placebo groups and were included in the analysis: RADIANT‐3 [14], RADIANT‐4 [15], SU‐1111 [13], and PROMID [10].

Table 3. Estimated OS in placebo groups at months 6, 12, 24, 36, 48, and 60.

Data are presented as estimated OS, % (95% CI).

Abbreviations: CI, confidence interval; N/A, data not available/not applicable; OS, overall survival.

Summary of Results for Placebo and Treatment Groups

Tables 4 and 5 and Figures 1 and 2 summarize the comparative pooled results between the placebo and active treatment arms.

Table 4. Meta‐analysis of progression‐free survival: comparison of pooled results between placebo and treatment arms.

Abbreviations: CI, confidence interval; PFS, progression‐free survival.

Table 5. Overall survival: comparison of pooled results between placebo and treatment arms.

Abbreviations: CI, confidence interval; OS, overall survival.

Figure 1.

Progression‐free survival: comparison of pooled results between placebo and treatment arms with 95% confidence intervals.

Figure 2.

Overall survival: comparison of pooled results between placebo and treatment arms with 95% confidence intervals.

Discussion

The meta‐analysis presented here reports pooled outcomes of six randomized clinical trials [8], [9], [10], [12], [13], [14], [15], [16] in which an active treatment arm (lanreotide, octreotide, everolimus, sunitinib, or Lu‐Dotatate) was compared with placebo in patients with metastatic NETs. The main aim was to assess progression‐free and overall survival outcomes among these heterogenous trials to determine if the reported benefits of active treatment versus placebo were maintained across a wide range of patient profiles and diverse medical treatments. Despite the wide heterogeneity of these trials, the current pooled analysis shows that active treatment improved survival outcomes—both PFS and OS—compared with placebo at all time points. The improved OS outcomes in the treatment arm are especially relevant given that the slow‐growing nature of these neoplasms—together with the crossover design of most trials—makes it difficult to determine the true impact of treatment on overall survival.

Progression‐Free Survival

All of the individual trials included in this meta‐analysis reported a significant benefit for active treatment for PFS outcomes (Table 4). The pooled data confirm this advantage at all assessed time points (3, 6, 12, 18, and 24 months). Although differences in the pooled PFS at month 3 were relatively modest (93% vs. 97%), these differences became greater over time, with a nearly 30% difference between the placebo and treatment arms at month 6 (54.3% vs. 83.7%). Interestingly, the greatest difference was observed at month 24, where the PFS for placebo was only 18% versus 61% for active treatment (a difference of >43 percentage points). Figure 1 depicts these differences graphically, showing the clear advantage for active treatment versus placebo, which increased over time. Although these data show a clear benefit for active treatment, they also reveal that a substantial proportion of patients are likely to remain progression free regardless of whether they receive treatment, a finding that is consistent with the CLARINET data [8]. However, because of a lack of reliable prognostic indicators, it is not currently possible to differentiate between these patients prior to making the treatment decision.

Overall Survival

There is clear evidence that SSAs, everolimus, Lu‐Dotatate, and sunitinib are all efficacious in slowing disease progression in patients with NETs; however, because of the long life expectancy of these patients, together with the fact that most of the trials [21] performed to date have allowed crossover from placebo to active treatment or salvage therapy, it is not known whether these treatments provide a true benefit in terms of OS. For this reason, PFS may be a more appropriate endpoint to assess differences in disease activity and as a surrogate marker for improved OS [22].

The present meta‐analysis demonstrates an advantage in OS for the treatment arms at all time points (Table 5; Fig. 2). Although this advantage was modest at months 6 and 12, the difference between placebo and active treatment increased at months 24 and 36 and especially at month 48 (49.9% vs. 61%). However, from month 48 to month 60, although active treatment maintained a survival advantage, the difference between placebo and active treatment was only 4.2 percentage points (41.7% vs. 45.9%). Despite these fluctuations, which may be attributable to the small sample sizes in these studies, the active treatment arm had better outcomes at all time points. Clearly, these findings must be interpreted cautiously given the reduced sample sizes of these RCTs and the methodological limitations of the present meta‐analysis. Nonetheless, these findings suggest that patients who receive active treatment may live longer than those who receive placebo or no treatment, thus raising doubts about the advisability of the active surveillance approach in this patient population.

Active Surveillance

Clinical support for active surveillance in patients with asymptomatic, low‐grade, nonfunctional tumors is based on the observation that a considerable proportion of patients do not progress for months or years (in some cases). The findings of this meta‐analysis confirm that a substantial percentage of patients remain progression free for an extended period, as evidenced by the pooled PFS at month 24 in the placebo groups (Table 4): 18% of patients who received placebo showed no disease progression, even after 2 years of follow‐up. The results for the placebo arm of the CLARINET trial were even better—probably because of the favorable prognostic characteristics of the study population—with close to one third of patients presenting no evidence of disease progression at 2 years.

It seems clear, therefore, that there is a subset of patients in which treatment can be delayed until there is evidence of disease progression. Nevertheless, even if the true 2‐year PFS is around 20% (as our data seem to indicate), the vast majority (80% or more) of patients who receive placebo will develop disease progression within 2 years. This finding underscores the need to distinguish between patients who require immediate treatment and those who may be candidates for active surveillance; unfortunately, the lack of reliable prognostic makers [14]—particularly in patients with low‐grade disease—makes it difficult to differentiate a priori between these patients. For this reason, until more reliable markers become available, the current evidence base—including the findings of this meta‐analysis—suggests that most patients, especially those with well‐differentiated, slow‐growing, somatostatin‐receptor‐positive tumors, should receive first‐line treatment with SSAs, particularly in light of the proven antiproliferative activity and the minimal adverse effects of these drugs [10], [11], [23].

Adverse Effects of SSAs Compared with Other Therapies

Numerous studies have shown that SSAs are well tolerated, with only mild to moderate adverse effects (AEs) in most patients [8], [24], [25]. In the CLARINET trial, the adverse event rate in the treatment and placebo arms was similar, and most treatment‐related AEs were either mild or moderate. Moreover, quality of life did not differ significantly between the treatment and placebo groups.

The relatively mild adverse‐event profile of SSAs compares favorably with other medical treatments for metastatic NETs. For example, in the RADIANT‐3 trial [14], although drug‐related (everolimus) AEs were mostly grade 1 or 2—including stomatitis (64% of patients), rash (49%), diarrhea (34%), and infections (23%)—severe grade 3 or 4 AEs (anemia and hyperglycemia) were observed in more than 5% of patients. Similarly, sunitinib malate has also been associated with severe AEs, including heart failure [12]. In the SU‐1111 study, dose interruptions were reported in 30% of the active treatment arm (vs. only 12% of the placebo group), primarily due to AEs such as neutropenia (12% of patients) and diarrhea (10%).

Spontaneous Tumor Regression in Patients Treated with Placebo

Recently, two meta‐analyses assessed tumor regression in the placebo arm of clinical trials of NETs [26], [27]. Amoroso et al. [26] performed a meta‐analysis of all prospective RCTs in which an active experimental treatment was compared with a placebo control arm, finding tumor shrinkage >10% from baseline in a subset of patients diagnosed with neuroendocrine tumors but also in 6% of placebo patients. These findings are relevant to the present study, in which approximately 18% of patients in the placebo arm remained progression free even after 24 months. These are clearly the patients who are suitable for active surveillance, but—as Amoroso and colleagues point out—the biological and clinical characteristics of these patients remain undefined. Ghatalia et al. [27] conducted a similar meta‐analysis, but in this case to investigate regression of advanced solid tumors in the placebo (or no anticancer therapy) arm. Notably, that meta‐analysis included two of the trials [12], [20] that were also included in our meta‐analysis. Both of the aforementioned meta‐analyses found that spontaneous remissions are common across all tumors, again underscoring the need to identify the characteristics of patients who may not require immediate treatment.

Treatment Indication and Selection

The PROMID and CLARINET trials, both of which demonstrated conclusively that SSAs can inhibit tumor growth and delay time to progression, have altered the clinical approach in this patient population. The data from those two studies, particularly the significantly delayed time to progression in the treatment group, strongly suggest a benefit for active treatment versus placebo. Early treatment administration is further supported by the proven benefits of SSAs (even in nonprogressors) and the well‐tolerated AE profile.

Although the decision to delay treatment in asymptomatic patients is understandable, exploratory data from the CLARINET trial on the tumor growth rate (TGR) demonstrate that even though some patients present a slow TGR, these tumors nonetheless continue to grow over time [28]. That exploratory analysis compared lanreotide with placebo, finding that lanreotide significantly reduced mean TGR after 12 weeks of treatment, a difference that was maintained at all subsequent follow‐up visits. Moreover, although the tumors continued to grow in both groups, they did so at a significantly lower rate in the lanreotide group.

In a recent editorial, Fazio [6] observed that the active surveillance approach has long been applied in patients with NETs because these tumors can remain stable for an extended period of time in some patients. Although active surveillance is recommended in clinical guidelines and other publications [5], the validity of this approach in these patients has never been confirmed. In this regard, we agree with Fazio that the current evidence base is insufficient to justify active surveillance, especially considering the tendency for NETs—even those with favorable biological characteristics—to continue growing. Support for active surveillance is further undermined by the proven effectiveness of SSAs in controlling symptoms and in improving PFS (and probably OS) with only minimal toxicity.

Although active surveillance can be considered an alternative to SSAs in some tumor localizations—primarily midgut and pancreatic NETs—the outcomes of the PROMID and CLARINET trials in midgut NETs [9] and in enteropancreatic tumors [8] make it difficult to justify this approach. In this regard, the retrospective study by Panzuto et al. in advanced pancreatic NETs [29] provides additional support for active treatment based on the significantly shorter PFS in untreated patients versus the treatment arm. The CLARINET trial shows that SSAs can be used for tumor control in a variety of neuroendocrine tumors, not just midgut NETs. The findings of the present meta‐analysis, which evaluated a wide range of treatments and localizations for metastatic NETs, would seem to support that conclusion.

Although active surveillance is not, on the whole, supported by the evidence, deferred treatment may be appropriate for certain well‐selected patients. Selecting the most appropriate treatment for these patients is often difficult because of the need to take into account a wide range of variables such as the location and characteristics of the primary tumor—which would include functionality, growth rate, tumor grade, and overall tumor burden—as well as patient preferences [5]. Treatment selection will also depend on the expertise and access to alternative treatments at the treating center.

Study Strengths and Limitations

The main limitation of this meta‐analysis is the study design, in which survival rates were estimated from survival curves obtained from the RCTs included in the study. As a result of this methodological approach, it was not possible to determine whether the observed differences were statistically significant. In addition, because this study included pooled data from different treatment types, the influence of the timing of treatment initiation on PFS outcomes could not be assessed. Moreover, the relatively small sample sizes in the RCTs included in this meta‐analysis limit the strength of our findings. Nonetheless, this is the first meta‐analysis to be performed to date in this patient population.

Conclusion

The findings of this meta‐analysis are consistent with previous reports, including the landmark CLARINET and PROMID trials. Although the study design does not permit us to draw any definitive conclusions, the pooled data from these randomized clinical trials suggest that active treatment was superior to placebo at all time points for both progression‐free and overall survival. These findings suggest that most patients could potentially benefit from active treatment, although more definitive data are needed to confirm this. Future studies should seek to identify prognostic factors to identify candidates for deferred treatment versus those who would benefit from early treatment initiation.

Author Contributions

Conception/design: Jaume Capdevila, Enrique Grande

Provision of study material or patients: Jaume Capdevila, Jorge Hernando, Alejandro Roman‐Gonzalez, Enrique Grande

Collection and/or assembly of data: Jaume Capdevila, Jorge Hernando, Santiago Perez‐Hoyos, Alejandro Roman‐Gonzalez, Enrique Grande

Data analysis and interpretation: Jaume Capdevila, Santiago Perez‐Hoyos, Enrique Grande

Manuscript writing: Jaume Capdevila, Jorge Hernando, Enrique Grande

Final approval of manuscript: Jaume Capdevila, Jorge Hernando, Santiago Perez‐Hoyos, Alejandro Roman‐Gonzalez, Enrique Grande

Disclosures

Jaume Capdevila: Novartis, Pfizer, Ipsen, Exelixis, Bayer, Eisai, Advanced Accelerator Applications, Amgen, Sanofi, Merck Serono (H), Eisai, Novartis, Ipsen, AstraZeneca, Pfizer, Advanced Accelerator Applications (RF); Jorge Hernando: Eisai, Ipsen, Pfizer (C/A); Alejandro Roman‐Gonzalez: Amgen, Sanofi, Novartis, Novo Nordisk, Valentech, Baxalta, Bayer (H), Novartis, Sanofi (SAB), Novartis, Sanofi, Bayer, Novo Nordisk (C/A); Enrique Grande: Pfizer, Novartis, Ipsen, Advanced Accelerator Applications, Lexicon (C/A, SAB). The other author indicated no financial relationships.

(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board