Phase II trial demonstrates the efficacy and safety of individualized, dosimetry-based ¹⁷⁷Lu-DOTATATE treatment of NET patients

Peptide receptor radionuclide therapy (PRRT) using ¹⁷⁷Lu-DOTATATE is a valuable treatment option for patients with somatostatin receptor–positive neuroendocrine tumors (NETs). A somatostatin analog (octreotate) coupled to the radionuclide ¹⁷⁷Lu binds to somatostatin receptors overexpressed on tumor cells, thus delivering ionizing radiation in a molecularly targeted radiotherapy approach. The treatment effect is due to DNA damage by the beta-particles emitted from ¹⁷⁷Lu, whereas gamma radiation enables imaging for uptake mapping and dosimetry.

PRRT is endorsed by the major neuroendocrine societies [1‐3] and approved for treatment of gastroenteropancreatic tumors (GEPNETs) (by the European Medicines Agency and the US Food and Drug Administration). The evidence of the efficacy and safety of ¹⁷⁷Lu-DOTATATE was long based on retrospective and single-arm studies [4, 5]. In 2017, this evidence was complemented with the results from the NETTER-1 trial, a randomized phase III trial where the superiority of ¹⁷⁷Lu-DOTATATE over somatostatin analogs (SSAs) was demonstrated for small intestinal NETs (SiNETs), the largest subgroup of GEPNETs [6].

The approved treatment approach of using a fixed activity of 7.4 GBq ¹⁷⁷Lu-DOTATATE in four cycles is safe but not necessarily the most effective treatment for individual patients, nor is it in line with current European legislation [7]. In radiotherapy, dose–response relationships are proven for most tumors [8], and the aim is to deliver a sufficiently high dose to the tumor with acceptable exposure to the organs at risk. The same type of dose–response/dose-toxicity relationships can be applied to radionuclide therapy, as is being confirmed by a growing body of evidence through the use of image-based dosimetry in prospective clinical trials [9‐14].

The main organs at risk are the kidneys and bone marrow. While bone marrow toxicity presents early and is therefore readily detected, renal toxicity occurs months to years after treatment. The exposure of the kidneys in PRRT is due to an active reabsorption of ¹⁷⁷Lu-DOTATATE in the renal proximal tubules [15], which can be partially inhibited by a parallel infusion of amino acids [16]. Previous experience with PRRT using ⁹⁰Y has demonstrated dose-dependent renal toxicity at a biologically effective dose (BED) of 28 Gy for patients with risk factors for renal toxicity and 40 Gy for those without risk factors [17]. A renal BED above 45 Gy was correlated with a high risk of rapid decline in renal function [18]. The radiation exposure of the kidneys therefore needs to be monitored. Carrying out as many treatment cycles as possible within a predefined dose limit to the kidneys is a first step towards dosimetry-based, individualized PRRT.

To improve the efficacy of ¹⁷⁷Lu-DOTATATE, the concept of an individually optimized treatment strategy needs to be further pursued and could be developed by adjusting the injected activity, modifying the treatment intervals, or increasing the total number of cycles. The latter strategy was assessed in the present phase II trial (ILUMINET), where the safety and efficacy of individualized ¹⁷⁷Lu-DOTATATE treatment based on the estimated renal BED were evaluated. Here we present the final results.

The results of this trial show that dosimetry-based PRRT for NET patients is safe and effective. Direct comparisons to the standard treatment used in the NETTER-1 trial are difficult since there are significant differences in the patient populations and trial designs, e.g., the proportion of low-grade NETs (2/3 in NETTER-1 and 1/3 in ILUMINET) and the time to progression prior to trial entry (36 months in NETTER-1 and 14 months in ILUMINET). Despite the higher risk population in ILUMINET, the best overall response rate of 34% compares favorably and the PFS and OS of 29 and 47 months repectively is simular to the results of NETTER-1 (18%, 28 and 48 months, respectively) [23].

Dosimetry-based PRRT has been conducted and reported by two other groups. The Uppsala observational study [24] had a design similar to ILUMINET, but the patient population included high-grade tumors. The number of cycles they administered was 1–10, median not reported, with 48% receiving more than four cycles. They reported a PFS of 27 months, with a response rate of 24%.

A Canadian trial used the same renal dose limit as the Uppsala study (23 Gy), but instead of varying the number of treatment cycles, they adjusted the injected activity in such a way that the renal dose limit was reached in four cycles [25]. This group reported a high response rate of 59% but a surprisingly short PFS of 16 months, and a high rate of, mainly hematological, grade 3–4 toxicity. This raises the question of the importance of the fractionation schedule—is it better to give smaller, repeated doses over a longer period of time than to give the same amount of activity over a shorter period? In the current study, the total median injected activity/patient was 37 GBq, compared to 36 GBq in the study by Del Prete et al. The duration of treatment was up to 97 weeks (median 44) in the current study, and up to 30 weeks in the latter. So, even though the protocol-specified dose constraint to the kidneys differed between the two trials, the total injected activity/patient was similar. There was, however, a large difference in the duration of the treatment period which could theoretically be a reason for the large difference in PFS. In our own material, there is a strong correlation (R² = 0.9, p < 0.01, data not shown) between cumulative BED and duration of treatment period, so we are not able to differentiate between these two potential explanatory factors for PFS.

The significant differences in PFS when subgrouped by Ki 67, ECOG, and renal BED are challenging to interpret since there are obvious confounders (such as treatment-related toxicity, prognosis, and selection bias). The fact that there was very limited toxicity in the highest BED group, and the poorer results for the patients with the highest Ki 67, would support a strategy in future trials to intensify treatment for the higher grade tumors with the hope of improving outcomes. No significant differences in PFS or OS were seen for different origins; however, pNET patients had a higher Ki 67% than SiNET patients (Table 3). This result has also been reported by others [4] and may reflect that pNETs are more responsive to radiation therapy in general than the more indolent group of SiNETs.

The results of the analysis of PFS and OS by renal BED beg the question of whether more cycles of treatment result in a higher dose to the tumor and thereby a better and more durable response. This question cannot be definitively answered in a nonrandomized setting, but the fact that the time to maximum response also increased by BED level from 12 months (BED < 25 Gy) to 23 and 28 months (BED 25–29 and > 29 Gy, respectively) implies that with increasing number of cycles the tumor shrinkage continues for a longer time.

When designing the trial, we expected a larger proportion of patients to be able to continue treatment in step 2. Sixty patients (62%) were excluded from step 2 because of earlier chemotherapy or liver-directed treatment (Table 1), and 7 patients (7%) were excluded solely based on age > 70 years. Given the limited toxicity noted in this trial, it seems safe to offer treatment beyond the proposed renal dose limits with less restrictive criteria, with the goal of a better long-term effect and survival.

When designing the trial, it had to be decided which renal dose limits to use based on existing evidence. The commonly used renal absorbed dose limit of 23 Gy originates from external beam radiotherapy (EBRT) and is based on delivering 2 Gy fractions at a dose rate of approximately 1 Gy/min, which has a very different biological effect than the dose rate achieved with PRRT. Assuming that the linear-quadratic model is valid for both EBRT and PRRT, 23 Gy given in 2 Gy fractions corresponds to approximately 40 Gy BED, while an absorbed dose of 23 Gy at the typical dose rate of PRRT corresponds to approximately 25 Gy BED (4.5 Gy/cycle, \(\alpha /\beta =2.6\), and repair half-life of 2.8 h assumed) [20]. Furthermore, the renal BED thresholds of 27 and 40 Gy used in the current trial may have been overly conservative, since ⁹⁰Y-PRRT has a more uniform energy deposition pattern than ¹⁷⁷Lu [26], possibly explaining the more pronounced renal toxicity. Renal function did seem to decline slightly during follow-up, however, which will be further studied in an ad hoc analysis with the goal of determining the relevant renal dose limit for ¹⁷⁷Lu-PRRT. An important step towards making dosimetry-based treatment generally available is to simplify the imaging procedure on which the dosimetry is based. Single or dual time-point SPECT imaging has been explored by our group and others and found to give reliable dosimetric estimates for the kidneys, especially when the last imaging time-point is at least 72 h after therapy [27‐31].

The biological effects of the absorbed radiation dose on tumor lesions and normal tissue in PRRT are likely influenced by several factors governed both by patient (size, renal function, risk factors for toxicity, etc.) and tumor (volume, proliferation rate, intrinsic radiation sensitivity) characteristics. To achieve a truly personalized PRRT, we need to understand which of them have the greatest impact on treatment outcomes and design the treatment accordingly [32]. Regardless of strategy, a personalized approach motivates dosimetry evaluations of organs at risk and ideally of the tumor as well. There are several situations when this information would facilitate treatment decisions, e.g., retreatment with ¹⁷⁷Lu-DOTATATE, treatment of patients with reduced kidney function or bone marrow reserve, and intensified treatment to more rapidly progressing tumors, e.g., NET G3 with a Ki 67 labeling index > 20%.

For many patients, the standard regimen of four cycles of ¹⁷⁷Lu-DOTATATE will provide valuable responses and survival benefits. The current results confirm that by individualizing the therapeutic approach the response rate can increase considerably, without causing clinically significant acute or late toxicity. This raises the question whether it is still justifiable to treat according to the currently approved, non-individualized regime. It is possible that it leads to a significant proportion of patients being undertreated, exposing them to the much greater threat to their survival than low-grade toxicity, namely tumor progression. Further improvement of the individualized treatment approach may be achieved by offering patients with more aggressive tumors higher intensity treatment, including combinations of PRRT with other systemic anti-cancer therapies, while patients with low-grade tumors continue to receive a therapy with minimal toxicity and impact on their everyday life. Thus, there is currently a high unmet need for well-designed randomized trials in PRRT, through which the risks and benefits of personalized vs. standard therapy can be demonstrated and quantified.