The Prognostic Value of the De Ritis Ratio for Progression-Free Survival in Patients with NET Undergoing [177Lu]Lu-DOTATOC-PRRT: A Retrospective Analysis

Simple Summary Peptide receptor radionuclide therapy (PRRT) of neuroendocrine tumors (NET) has shown variable response rates between 9% and 39%. Therefore, better criteria are needed that help doctors to identify patients who will show a favorable outcome to PRRT, and which patients may not. The so-called De Ritis ratio, which is calculated using two basic laboratory parameters of liver function, has shown that it can help to predict the patient outcome in various tumor types. This retrospective study included 125 patients with NET who were treated with PRRT. We demonstrated that a high De Ritis ratio and high levels of the tumor marker Chromogranin A (CgA) each improved the prediction of the progression-free survival after treatment. A consequence for clinical care might be that patients with both high De Ritis ratio and high CgA levels may benefit from intensified follow-up imaging after PRRT because they have a higher risk of early progression. Abstract Background: The De Ritis ratio (aspartate aminotransferase [AST]/alanine aminotransferase [ALT]) has demonstrated prognostic value in various cancer entities. We evaluated the prognostic capability of the De Ritis ratio in patients with metastatic neuroendocrine tumors (NET) undergoing peptide receptor radionuclide therapy (PRRT). Methods: Unicentric, retrospective analysis of 125 patients with NET undergoing PRRT with [177Lu]Lu-DOTATOC (female: 37%; median age: 66 years; G1+G2 NET: 95%). The prognostic value regarding progression-free survival (PFS) was analyzed with univariable and multivariable Cox regression. Prognostic accuracy was determined with Harrell’s C index and a likelihood ratio test. Results: Progression, relapse, or death after PRRT was observed in 102/125 patients. Median progression-free survival (PFS) was 15.8 months. Pancreatic or pulmonary origin, high De Ritis ratio, and high Chromogranin A (CgA) significantly predicted shorter PFS in univariable Cox. In multivariable Cox regression, only high De Ritis ratio >0.927 (HR: 1.7; p = 0.047) and high CgA >twice the upper normal limit (HR: 2.1; p = 0.005) remained independent predictors of shorter PFS. Adding the De Ritis ratio to the multivariable Cox model (age, Eastern Cooperative Oncology Group (ECOG) performance status, primary origin, CgA) significantly improved prognostic accuracy (p < 0.001). Conclusions: The De Ritis ratio is simple to obtain in clinical routine and can provide independent prognostic value for PFS in patients with NET undergoing PRRT.


Introduction
Neuroendocrine tumors (NET) of the bronchopulmonary (lung-) or gastroenteropancreatic system (GEP-NET) represent a rare and heterogeneous class of tumors [1]. During the last few decades, the incidence of NET has significantly increased more than six-fold, from 1.1/100,000 persons to 7.0/100,000 persons, mainly due to the progress in functional imaging, device-specific sensitivity, and further increasing awareness of the occurrence of NET [1,2].
For metastatic well-differentiated GEP-NET of low and intermediate grade G1 and G2, peptide receptor radio nuclide therapy (PRRT) with [ 177 Lu]Lu-[DOTA 0 -Tyr 3 ]octreotate ([ 177 Lu]DOTATOC) is a well-established second-to third-line therapy after progress under treatment with common "cold" somatostatin analogues [3]. The NETTER-1 study, the first randomized controlled trial in patients with metastasized GEP-NET and treatment with PRRT, showed an average therapy response rate of 18% in these patients [4]. However, prognosis varies, and not all patients benefit from PRRT. Many retrospective studies found response rates between 9 and 39% [5,6]. A recent meta-analysis including 11 studies and 1268 patients reported an average response rate of 29.1% [7]. Therefore, better stratification criteria are highly desirable to identify patients who will ultimately show a favorable response and longer progression-free survival (PFS) after PRRT.
The ratio of aspartate aminotransferase [AST]/alanine aminotransferase [ALT] in the pretherapeutic blood serum or heparin plasma, the so-called ''De Ritis ratio" [20], has been recently reported to be a valuable and independent prognostic factor in the treatment of different tumor entities [21][22][23][24][25][26][27][28][29]. Furthermore, a representative meta-analysis, which included 9400 patients with different types of cancer, demonstrated that a high AST/ALT ratio was associated with an impaired overall survival (OS) [30]. Until now, the De Ritis ratio has not been addressed in the context of NETs. The primary aim of this study was to evaluate the prognostic value of the pretherapeutic De Ritis ratio on the PFS in patients with NET undergoing PRRT. Secondly, a combined risk score of high De Ritis ratio and high CgA was evaluated to predict the PFS.

Predictive Model for Progression-Free Survival
The Cox model, including high age, ECOG score, primary tumor location, and high CgA (model 1), showed a likelihood ratio (LR) χ 2 of 11.5 and Harrell's C of 0.633. The predictive accuracy was slightly but significantly higher if the Cox model included high age, ECOG score, primary tumor location and the combined score of high CgA and high De Ritis ratio (model 2; LR χ 2 = 14.0, Harrell's C = 0.65, LR test: p < 0.001).

Predictive Model for Progression-Free Survival
The Cox model, including high age, ECOG score, primary tumor location, and hi CgA (model 1), showed a likelihood ratio (LR) χ 2 of 11.5 and Harrell's C of 0.633. T predictive accuracy was slightly but significantly higher if the Cox model included hi age, ECOG score, primary tumor location and the combined score of high CgA and hi De Ritis ratio (model 2; LR χ 2 = 14.0, Harrell's C = 0.65, LR test: p < 0.001). Combining

Discussion
The aim of this study was to assess the prognostic value of the pretherapeutic D Ritis ratio regarding PFS in patients with NET undergoing treatment with [ 177 Lu]Lu-D TATOC PRRT. To the best of our knowledge, there has not been a previous study analy ing the prognostic value of the De Ritis ratio in the context of NET. However, Wu et

Discussion
The aim of this study was to assess the prognostic value of the pretherapeutic De Ritis ratio regarding PFS in patients with NET undergoing treatment with [ 177 Lu]Lu-DOTATOC PRRT. To the best of our knowledge, there has not been a previous study analyzing the prognostic value of the De Ritis ratio in the context of NET. However, Wu et al. included 8853 patients with various tumor entities in a meta-analysis and illustrated the additional value of the De Ritis ratio as a prognostic parameter for survival outcomes (OS: HR = 1.7, p < 0.001) [30].
At present, a compelling explanation of the prognostic value of the De Ritis ratio remains to be found [20]. Both enzymes AST and ALT are routinely determined as parts of the "liver function panel" [31]. The physiologic AST/ALT ratio in hepatocytes is 2.5/1, while ALT presents a two-fold increased biological half-life (t 1 2 = 36 h) in comparison to AST (t 1 2 = 18 h) [31]. Therefore, the concentration of these enzymes should be equalized in the blood over the long term [31]. However, in the case of an increased rate of hepatocyte apoptosis, the serum ratio of AST/ALT (De Ritis ratio) rises as more AST and ALT are released than eliminated according to the half-life [31,32]. Nonetheless, it should be noted that AST is expressed in various tissue types, while ALT is more liver-specific [31]. Consequently, an increased AST is not always caused by hepatic pathology, especially if ALT is not comparably and simultaneously elevated [31,33]. It is currently hypothesized that anaerobic glycolysis, which is typical of the metabolism of cancer cells (the "Warburg effect") may explain the prognostic value of the De Ritis ratio [21][22][23][24][26][27][28][29][30]34,35]. In this context, several interactions exist between increased anaerobic glycolysis, an altered NAD+/NADH ratio in the cytoplasm, and AST, which is essential for the function of the malate-aspartate shuttle [31,[36][37][38]. Underlining this hypothesis, Thornburg et al. demonstrated that cancer cells are especially dependent on AST for a high proliferation rate [39]. In the current sample, we therefore suspect that the elevated De Ritis ratio in the corresponding subgroup was more likely related to a higher tumor proliferation rate than to a hepatic source, as ALT was significantly lower in this subgroup.
Additionally, we could report that a high pretherapeutic CgA was a significant predictor of shorter PFS in multivariable Cox regression. Previously, several authors could demonstrate the prognostic value of CgA in patients with NET undergoing treatment with PRRT [10,12,13]. In a recent analysis by Aalbersberg et al., patients with a CgA ≥336 µg/L (median) had a significantly shorter PFS in multivariable Cox regression than patients with a CgA <112 µg/L (first quartile) [10]. These studies are well in line with our finding, which found that patients demonstrated a significantly shorter PFS with CgA values > twice the upper normal limit (2xULN). The median of our analysis, nevertheless, was almost two times higher (336 vs. 612 µg/L). However, we used a different cut-off (2xULN) as this cut-Cancers 2021, 13, 635 8 of 12 off was already validated in a prospective, multinational, phase 2 study [40]. Differences may arise, since Aalbersberg et al. only included patients who received at least 3 cycles of PRRT [10]. CgA values should be generally compared cautiously since they may differ between laboratories [41].
This study was limited by its retrospective nature and lack of a matching control group of patients undergoing a different treatment. Therefore, a predictive capability of the De Ritis ratio could not be formally assessed. Prospective studies are required to validate the current explorative results and to ensure a well-selected, homogenous patient collective. Moreover, due to the retrospective setting, we could not assess descriptive follow-up parameters of laboratory values or imaging findings beyond clinical routine data, which should be addressed in subsequent prospective studies. As recently demonstrated, values for AST, ALT, and CgA might increase after administration of PRRT cycles [42,43]. Given the retrospective nature of this study design, data on the short-term dynamic of these laboratory values were not available to detect these reversible alterations. Therefore, prospective evaluation with close post-therapeutic monitoring of the parameters may be useful to investigate the prognostic relevance of their post-therapeutic dynamic. Notably, this would be independent from the current observation of their prognostic value prior to treatment. Furthermore, patients with a high De Ritis ratio were, on average, significantly older and showed significantly higher CgA values (Table 1). In principle, this might introduce a bias in evaluating a prognostic value of the De Ritis ratio. However, multivariable Cox regression demonstrated that an independent prognostic value of the De Ritis ratio remained after adjustment for these other risk factors. A multivariable Cox regression in an unmatched cohort was favored over a matched-pair analysis, e.g., using propensity score matching as conducted by other groups [27,29]. As reported by Biondi-Zoccai et al., multivariable Cox regression should be preferred over propensity score matching if the ratio of events to variables is >8-10 (as in the current analysis), because matching usually involves discarding patients without a proper match. This reduces statistical power [44], which other researchers have confirmed [45].
Determination of both CgA and De Ritis ratio is inexpensive and already part of a clinical routine in patients with NET. Therefore, the necessary validation of the current results in prospective (multicenter) studies should be straightforward. Moreover, proof of a predictive value of De Ritis ratio, in addition to its prognostic significance [46], will be required, as only this will have direct impact on clinical decisions among a broad spectrum of therapeutic strategies for patients with well-differentiated NET.

[ 177 Lu]Lu-DOTATOC-PRRT and Response Assessment
Patients underwent PRRT with a median of 3 cycles (range: 1-6 cycles) and a scheduled dose of 200 mCi (7.45 GBq) [ 177 Lu]Lu-DOTATOC per cycle. PRRT cycles were administered in intervals of 10 to 12 weeks. After application of two cycles of PRRT all patients underwent SSR-PET/CT with [ 68 Ga]Ga-DOTATOC as interim staging for response evaluation, which was repeated every two cycles. In addition, interim staging was generally performed at least 2 months after application of the last cycle PRRT to avoid misinterpretation due to possible pseudo-progression (radiogenic edema) [42]. Progressive disease was determined by an interdisciplinary tumor board. In the case of progressive disease, no further PRRT cycles were administered. After treatment completion, patients underwent follow-up imaging every 3 to 6 months. Morphological assessment was generally performed by contrast-enhanced (CE) CT. Alternatively, CE magnetic resonance imaging (MRI) was used for morphological evaluation if available.

Evaluation
AST, ALT, and CgA were determined <4 weeks before application of the first cycle of PRRT (De Ritis ratio = ALT/AST). PFS was defined as the time from the first cycle of PRRT until detection of progressive disease according to the response evaluation criteria in solid tumors (RECIST) 1.1 or death from any cause [47].

Statistical Analysis
Statistical analysis was performed using SPSS version 25 (IBM, Chicago, IL, USA) and R 4.0.0 (Foundation for Statistical Computing, Vienna, Austria, 2020; http://www.Rproject.org (accessed on 19 January 2020)). Significance was assumed at A= 0.05. Descriptive values were expressed as median, IQR, and range. Univariable Cox proportional hazards regression regarding the PFS included clinical parameters (sex, age, functionality of the NET, presence of a Hedinger syndrome, localization of the primary tumor (gastrointestinal, pancreatic, pulmonary, cancer of unknown primary), localization of metastases (hepatic, lymphonodal, osseous, peritoneal, pulmonary), ECOG score [48] and Charlson comorbidity index (CCI) [49]) and laboratory parameters (CgA, De Ritis ratio).The HR and the 95% CI were determined for each parameter. Before inclusion, continuous variables (age, CgA, De Ritis ratio) were binarized. The cut-off for age (>66 years) was defined as the median value in the patient sample, while the CgA cut-off (>204 µg/L) was 2xULN [40,50]. The De Ritis ratio was binarized with a cut-off (>0.927) that achieved the minimum p-value in the log-rank test as determined with the Charité Cutoff Finder [51]. Variables were compared between groups with low vs. high De Ritis ratio using the Wilcoxon rank-sum test (continuous variables) or Fisher's exact test (categorical variables). All variables with p ≤ 0.1 in univariable Cox regression were also candidates for inclusion into multivariable Cox regression. The proportional hazard assumption was tested using the goodness-of-fit test and fulfilled by each variable. Using equal weights, the factors high CgA and high De Ritis ratio were combined as a predictive model regarding PFS. Using the rms package for R, the LR χ 2 and Harrell's C index of the multivariable Cox model were calculated after inclusion of either high age, ECOG, primary tumor location, and high CgA (model 1), or high age, ECOG, primary tumor location, and the combined score of high CgA and high De Ritis ratio (model 2) [52]. The likelihood ratio test for these two Cox models was performed to test if the combined score of CgA and De Ritis ratio provides additional prognostic value over CgA alone. The Kaplan-Meier method was used to estimate survival rates and average PFS.

Conclusions
The De Ritis ratio provided an independent prognostic value for PFS in patients with NET undergoing PRRT with [ 177 Lu]Lu-DOTATOC. Consequently, the follow-up in patients with both a high De Ritis ratio and high CgA might be intensified as they have a higher risk for early progression. Its assessment as a routine clinical laboratory parameter is straightforward, while the cause of its prognostic value remains unclear, but likely relates to tumor metabolic activity instead of liver function, per se. Institutional Review Board Statement: The study was conducted in accordance with the guidelines of the Declaration of Helsinki, and approved by the Institutional Ethics Committee of the Charité-Universitätsmedizin Berlin (reference ID: EA2/006/20; vote 26 January 2020).

Informed Consent Statement:
Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.