DiM: Prognostic Score for Second- or Further-line Immunotherapy in Advanced Non–Small-Cell Lung Cancer: An External Validation

(1)

Original Study

DiM: Prognostic Score for Second- or Further-line

Immunotherapy in Advanced Non

eSmall-Cell

Lung Cancer: An External Validation

Arsela Prelaj,

1

Giuseppe Lo Russo,

1

Claudia Proto,

1

Diego Signorelli,

1

Roberto Ferrara,

1

Giulia Galli,

1

Alessandro De Toma,

1

Giovanni Randon,

1

Filippo Pagani,

1

Benedetta Trevisan,

1

Monica Ganzinelli,

1

Nicoletta Zilembo,

1

Michele Montrone,

2

Vito Longo,

2

Francesco Pesola,

2

Pamela Pizzutilo,

2

Gabriella Del Bene,

2

Niccolò Varesano,

2

Domenico Galetta,

2

Valter Torri,

3

Marina Chiara Garassino,

1

Massimo Di Maio,

4

Annamaria Catino

2

Abstract

No biomarkers, other than programmed cell death ligand 1, have been approved to guide the choice of immunotherapy for nonesmall-cell lung cancer. We evaluated whether the clinical prognostic score, DiM, could predict outcomes of patients with NSCLC receiving immunotherapy. DiM score could predict outcomes, especially for those with the worst prognosis.

Background: Other than the programmed cell death ligand 1 (PD-L1) value, oncologists have only the clinical characteristics of patients with advanced nonesmall-cell lung cancer (aNSCLC) to determine candidates for immu-notherapy. A clinical prognostic score composed of the Eastern Cooperative Oncology Group performance status, sex, histologic type, stage, platinum-basedfirst-line therapy, and response to first-line therapy has categorized 3 prognostic groups for patients undergoing second-line chemotherapy. We sought to validate the same score for patients with aNSCLC treated with second- or further-line immunotherapy. Materials and Methods: We collected data from 2 Italian centers. A score was generated to divide patients into 3 prognostic groups: best, score< 5; intermediate, score 5 to 9; and worst, score > 9. Overall survival (OS) and progression-free survival (PFS) were the endpoints. Results: A total of 347 patients were included for analysis. Their median age was 66 years (range, 30-88 years), most were aged< 70 years (67.5%), 70.7% were men, 79.5% were smokers, and 74.6% had had adenocarcinoma. The Eastern Cooperative Oncology Group performance status was 0 for 23%, 1 for 54.5%, and 2 for 22.5%. Of the 347 patients, 28% were in the best prognosis, 51% in the intermediate, and 21% in the worst prognosis group, respectively. The median OS was 18.0 months for the best, 8.5 months for the intermediate (hazard ratio [HR] vs. best, 1.83; 95% confidence interval [CI], 1.35-2.47; P < .001) and 2.6 months for worst (HR vs. best, 5.77; 95% CI, 3.99-8.33; P< .001) group. The median PFS was 3.4 months for the best, 3.7 months for the intermediate (HR vs. best, 1.35; 95% CI, 1.03-1.77; P ¼ .032), and 1.9 months for the worst (HR vs. best, 2.51; 95% CI, 1.80-3.50; P< .001) group. Conclusions: The prognostic score was able to predict the outcomes of patients with aNSCLC who had received immunotherapy. The worst category showed a dismal life expectancy

A.P. and G.L.R. contributed equally to the present study. 1

Department of Medical Oncology, Fondazione IRCCS Istituto Nazionale Tumori, Milan, Italy

2

Medical Thoracic Oncology Unit, IRCCS Oncologico Giovanni Paolo II of Bari, Bari, Italy

3_{Pharmacological Research Institute, IRCSS Mario Negri, Milan, Italy} 4_{Division of Medical Oncology, Department of Oncology, University of Turin,} Mauriziano Hospital, Turin, Italy

Submitted: Aug 11, 2019; Revised: Dec 21, 2019; Accepted: Jan 20, 2020

Address for correspondence: Arsela Prelaj, MD, Department of Medical Oncology, Fondazione IRCCS Istituto Nazionale Tumori, Via Giacomo Venezian, 1, Milan 20133, Italy

E-mail contact:arsela.prelaj@istitutotumori.mi.it

(2)

-and probably would not beneﬁt from active systemic therapy. Thus, for these patients, best supportive care could be the best choice.

Introduction

Immune checkpoint inhibitors (ICIs) have significantly changed the therapeutic landscape of advanced nonesmall-cell lung cancer (aNSCLC).1Currently, pembrolizumab is the standard of care as first-line therapy for programmed cell death ligand 1 (PD-L1) 50% NSCLC (KEYNOTE-024).2,3 Nivolumab and pembrolizumab were thefirst 2 ICIs approved on the basis of the significant improvement in overall survival (OS) compared with docetaxel in patients with pretreated aNSCLC, with both squamous and nonsquamous histologic features.4-6 _{Atezolizumab is another}

therapeutic option in the same setting,7and durvalumab demon-strated activity even in patients with strongly pretreated aNSCLC.8 Despite the improvement in survival, only a limited number of patients will respond to immunotherapy (IO) and even fewer pa-tients will experience a durable response.6 The progression-free survival (PFS) and OS curves in pivotal second-line (2L) studies with nivolumab, pembrolizumab, and atezolizumab showed an overlap in the ﬁrst months of therapy, demonstrating that most patients will not respond to ICIs, with a non-negligible risk of early clinical failure.4-8 The discovery of predictive and prognostic bio-markers continues to be of great interest.

PD-L1 expression is the only biomarker approved for use in aNSCLC; however, multiple studies have reported several biologic and technical limitations owing to the temporal changes in PD-L1 expression and its intratumor heterogeneity.9 _{Conﬂiction results}

were described for a proportion of patients who had responded to ICIs with low or negative PD-L1 expression6,10; thus, its negative predictive role has remained suboptimal.11 Other than PD-L1 expression, oncologists have only patients’ clinical characteristics to determine whether patients are candidates for ICIs therapy.12-15 Other potential biomarkers have recently been investigated, including the tumor mutation burden, immune score, CD8þ tumor-inﬁltrating lymphocytes, and the immune gene signature, with interesting results.16 However, to date, none of these factors has achieved a deﬁnite role in clinical practice.16

A crucial factor is also cancer-associated inﬂammation, which has been correlated with worse outcomes.17,18Moreover, different pe-ripheral blood parameters have been investigated in various malig-nancies.11,19_{Blood inﬂammatory biomarkers have been correlated}

with a low therapeutic response to conventional treatment and a poor prognosis17and have demonstrated an association with sur-vival outcomes in patients with advanced melanoma receiving ICIs.20,21Their prognostic role has also been reported in patients with aNSCLC treated with ICIs.22-24 However, few data exist regarding the predictive role of peripheral blood biomarkers in aNSCLC treated with antiePD-L1 inhibitors.22-28

A clinical prognostic score developed by Di Maio et al29(the DiM score) consists of the Eastern Cooperative Oncology

Group (ECOG) performance status (PS), patient sex, histologic type (ie, squamous, adenocarcinoma, other), stage (IIIB or IV), platinum-based first-line therapy, and response to first-line therapy (ie, complete or partial response, no response). The DiM score was used to categorize patients receiving second-line chemotherapy (CHT) into 3 prognostic groups.29The score was developed using individual data from 1197 patients enrolled in 9 randomized trials of 2L CHT. Subsequently, the DiM score was externally validated using data from 551 patients enrolled in a randomized phase III trial comparing vinflunine with doce-taxel in the same setting, confirming its prognostic importance.30

The aim of the present study was to determine whether the DiM score, developed and validated for patients receiving CHT, would be able to predict the outcomes of patients with aNSCLC receiving second- or further-line IO to identify those patients less likely to respond and to potentially help in the treatment choices.

Materials and Methods

The present study was conducted at the Fondazione IRCCS Istituto Nazionale Tumori of Milan and the IRCCS Oncologico Giovanni Paolo II of Bari, Italy. The present study was performed in accordance with the Declaration of Helsinki and the Good Clinical Practice and local ethical guidelines. All enrolled alive patients provided written informed consent.

Study Population

From August 2015 to December 2018, we conducted a retro-spective 2-center study of 347 patients with consecutive aNSCLC receiving single-agent antiePD-L1 inhibitors in 2L or further-line therapy (193 from Milan [C1] and 154 from Bari [C2]).

The eligible patients had fulﬁlled the following inclusion criteria: a cytologic or histologic diagnosis of aNSCLC and pretreated pa-tients (relapse or stage IIIB-IV) who had received 1 infusion of an antiePD-L1 agent as second- or further-line therapy. Patients who had received ICIs within a clinical study were also included. The exclusion criterion was the receipt of IO as ﬁrst-line therapy or combined with other systemic drugs. The clinical characteristics, including the treatment information from both centers, are reported inTable 1.

Treatment

ICIs were administered intravenously. Nivolumab was initially given at a dose of 3 mg/kg and, later, since May 2018, at a ﬁxed dose of 240 mg every 2 weeks. Pembrolizumab was given at a dose of 2 mg/kg every 3 weeks for patients with PD-L1 expression of 1%. Atezolizumab was given at a ﬁxed dose of 1200 mg every 3 weeks and durvalumab at a dose of

(3)

-Table 1 Patient Characteristics Stratiﬁed by Participating Institution and for Whole Series Characteristic C1 (n[ 193; 55.4%) C2 (n[ 154; 44.6%) C1 D C2 (n [ 347; 100%) P Value Sex <.001 Male 120 (62.2) 126 (82) 246 (70.7) Female 73 (37.8) 28 (18) 102 (29.3) Age, y .843 Median 65 67 66 Range 30-88 31-86 30-88 Age group, y .946 <70 131 (67.9) 104 (68.0) 235 (67.5) 70 62 (37.8) 50 (32.0) 113 (32.5) ECOG PS <.001 0 71 (36.8) 9 (5.8) 80 (23.0) 1 98 (50.7) 91 (59) 189 (54.5) 2 24 (12.5) 54 (35.2) 78 (22.5) Smoking status .347 Former/current smoker 150 (77.8) 126 (81.8) 276 (79.5) Never smoker 43 (22.2) 28 (18.2) 71 (20.5)

Smoking duration, pack-y

Median 30 50 40 Range 0-177 0-168 0-177 Histologic subtype _<.001 Adenocarcinoma 144 (74.6) 83 (53.9) 227 (65.4) Squamous 44 (22.8) 65 (42.2) 109 (31.5) Other 5 (2.6) 6 (3.9) 11 (3.1) EGFR/ALK alteration .384 Yes 13 (6.7) 7 (4.6) 20 (5.7) No 180 (93.3) 147 (95.4) 327 (94.3) Stage .015 III 5 (2.6) 13 (8.4) 18 (5.2) IV 188 (97.4) 141 (91.6) 329 (94.5) Liver metastases .961 Yes 38 (19.7) 30 (20) 68 (19.6) No 155 (80.3) 124 (80) 279 (80.4) Bone metastases .004 Yes 87 (45.1) 46 (30) 133 (38.3) No 106 (54.9) 108 (70) 214 (61.7) Brain metastases .019 Yes 44 (22.8) 20 (13) 64 (18.4) No 149 (77.2) 134 (87) 283 (81.6) Treatment line .174 Second line 118 (61.1) 105 (68) 223 (64.3)

Third line or more 75 (38.9) 49 (32) 124 (35.7)

ICI type .002 Anti_ePD-L1 33 (17) 10 (6.5) 43 (12.4) Anti_ePD-1 160 (83) 144 (93.5) 304 (87.4) First-line therapy .600 Platinum based 179 (92.7) 145 (94.2) 324 (93.3) Non-platinum based 14 (7.3) 9 (5.8) 23 (6.7) Radiotherapy .557 Yes 36 (18.6) 25 (16.2) 61 (17.6) No 157 (81.4) 129 (83.8) 286 (82.4)

(4)

-10 mg/kg every 2 weeks. Treatment was continued until disease progression, unacceptable toxicity, withdrawal, or death. Treat-ment after the occurrence of progressive disease (PD) was permitted if clinical beneﬁt would result according to the cli-nician’s judgment.

Statistical Analysis

The primary endpoint was OS and its association with the prognostic score to identify potential poor prognostic groups less likely to obtain a favorable outcome with ICIs. OS was calculated from the start of ICI treatment until death or the last follow-up examination.

The secondary endpoints were PFS and its association with the prognostic score. PFS was calculated from the date of theﬁrst ICI administration until PD, death from any cause, or the last follow-up visit for patients alive without PD. The Kaplan-Meier method was used to calculate the median PFS (mPFS) and median OS (mOS) with the 95% conﬁdence intervals (CIs) and to generate survival curves for PFS and OS according to the 3 risk categories. The median follow-up period was calculated using the inverted Kaplan-Meier technique.31 _{The log-rank test (Mantel-Cox) was}

used to evaluate statistically signiﬁcant differences in PFS and OS, which was deﬁned at the P < .05 level. The prognostic score was computed using the scoring system presented inTable 2. Similar to previous studies, patients were divided into 3 groups using cutoff scores well balanced along the range of values: best prog-nosis, score< 5; intermediate prognosis, score 5 to 9; and worst prognosis, score> 9. A Cox proportional hazard model was used to compare the 3 categories.32 The concordance C-index was calculated using the model proposed by Pencina and D’Agostino33

to measure the power of the discriminations. All statistical analyses were performed using the SPSS program, version 25.0 (IBM Corp, Armonk, NY) and R, version 3.5.1 (R Foundation, Vienna, Austria).

Response Evaluation

The radiologic assessments consisted of a baseline total body computed tomography scan. Computed tomography scans were subsequently performed every 3 to 4 cycles or whenever PD was clinically suspected.

The tumor response was assessed using the Response Evaluation Criteria in Solid Tumors, version 1.1.34The overall response rate (ORR) was defined as the sum of the complete and partial responses and the disease control rate as the sum of the complete and partial responses and stable disease. Patients with PD for whom clinical benefit would be maintained according to the clinician’s judgment could receive treatment after PD and were considered to have stable disease if PD had not been confirmed.

Results

Patient Characteristics

A total of 347 patients with aNSCLC who had received an antiePD-L1 inhibitor as 2L or further-line therapy were included in the present analysis. The patient characteristics are summarized in

Table 1. Of the 347 patients, 246 were men (70.7%) and 276 were smokers (79.5%). The median age was 66 years (range, 30-88 years), and 113 patients (32.5%) were aged> 70 years. The median ECOG PS was 1 (range, 0-1), with 22.5% patients having an ECOG PS of 2. The histologic subtype was adenocarcinoma in 65.4%, squamous cell in 31.5%, and other in 3.1%. At baseline, bone, liver, and brain metastases were present in 38.3%, 19.6%, and 18.4% of the patients, respectively. Two thirds of the patients (64.3%) had received ICIs as 2L therapy, and 124 patients had received antiePD-L1 inhibitors as third-line or further therapy.

General Response and OS Outcomes

All 347 patients included in the present study were assessable for the survival analysis. At the data cutoff point (December 2018), 306 patients (88%) had experienced PD and 260 patients had died

Table 1 Continued

Characteristic C1 (n[ 193; 55.4%) C2 (n[ 154; 44.6%) C1 D C2 (n [ 347; 100%) P Value

ORR toﬁrst-line therapy .031

Yes 61 (31.6) 66 (42.8) 127 (36.6) No 132 (68.4) 88 (57.2) 220 (63.4) Outcome with IO PFS, mo .005 Median 2.3 3.8 3.1 95% CI 1.9-2.6 3.3-4.3 2.7-3.5 OS, mo .761 Median 7.7 7.5 7.6 Range 5.4-9.9 4.4-10.6 5.7-9.5 ORR 34 (17.5) 22 (14.0) 56 (16.2) .402 95% CI 12.9-23.6 9.6-20.1 12.6-20.4 DCR 84 (43.5) 69 (44.8) 153 (44.1) .811 95% CI 36.7-50.5 37.2-52.7 39.0-49.4

Data presented as n (%), unless otherwise noted.

Abbreviations: C1¼ center 1 (Milan); C2 ¼ center 2 (Bari); CI ¼ conﬁdence interval; DCR ¼ disease control rate; ECOG ¼ Eastern Cooperative Oncology Group; ICI ¼ immune checkpoint inhibitor; IO¼ immunotherapy; ORR ¼ overall response rate; OS ¼ overall survival; PD-1 ¼ programmed cell death; PD-L1 ¼ programmed cell death ligand 1; PFS ¼ progression-free survival; PS ¼ performance status.

(5)

-(75%). Overall, after a median follow-up of 29 months (95% CI, 25.2-34.6 months), the mPFS was 3.1 months (95% CI, 2.6-3.5 months) and the mOS was 7.6 months (95% CI, 5.7-9.5 months). The ORR and disease control rate were 16.2% (95% CI, 12.6%-20.4%) and 44.1% (95% CI, 39.0%-49.4%), respectively.

Prognostic Index

The index score was assigned and calculated using the proposed scoring system from previous reports by Di Maio et al29,30 (Table 2). The patients were separated into 3 categories. Of the 347 patients, 96 (27.7%) had had a low score (< 5, best category), 178 (51.3%) had had a score of 5 to 9 (intermediate category), and 73 (21%) had had a high score (> 9, worst category).

Survival Results Among the 3 Groups

The mOS was 18.0 months (95% CI, 11.1-24.8 months) for the best group, 8.5 months (95% CI, 6.6-10.3 months) for the inter-mediate group, and 2.6 months (95% CI, 1.8-3.4 months) for the worst group. The Kaplan-Meier curves for the 3 prognostic groups are shown inFigure 1. We used the Cox hazard model to describe the differences between the intermediate and best prognosis groups (hazard ratio [HR], 1.83; 95% CI, 1.35-2.47; P< .001) and worst and best prognosis groups (HR, 5.77; 95% CI, 3.99-8.33; P < .001). The C-index of the model for OS was 0.67 (95% CI, 0.63-0.70).

As a secondary endpoint, we evaluated the PFS for the 3 cate-gories. The mPFS was 3.4 months (95% CI, 2.1-4.7 months), 3.7 months (95% CI, 3.2-4.2 months), and 1.9 months (95% CI, 1.5-2.2 months) for the best, intermediate, and worst category,

respectively. The Kaplan-Meier curves for PFS are shown in

Figure 2. The Cox model showed that the difference was statistically signiﬁcant between the intermediate and best groups (HR, 1.35; 95% CI, 1.03-1.77; P¼ .032) and worst and best groups (HR, 2.51; 95% CI, 1.80-3.50; P< .001).

Multivariate Analysis of OS

A multivariate analysis of the OS for the overall population was performed. The multivariate analysis included the following baseline patient characteristics: age, sex, smoking status, pack-years, ECOG PS, histologic type, stage, liver, brain, or bone metastases, use of ICIs as 2L or further-line therapy, platinum-based first-line therapy, and ORR at first-line therapy (Table 3). Only the ECOG PS (0 vs. 2: HR, 0.14; 95% CI, 0.092-0.213; P< .0001; 1 vs. 2: HR, 0.33; 95% CI, 0.23-0.45; P < .0001), presence of liver metastases at baseline (HR, 1.73; 95% CI, 1.27-2.36; P¼ .001), and smoking status expressed in pack-years (HR, 0.71; 95% CI, 0.53-0.96; P ¼ .026) were confirmed as relevant prognostic factors.

Results Stratifying the Model by Institution

To identify differences between the 2 institutions, we imple-mented separated survival analyses for OS and PFS for C1 and C2. The mOS was 7.6 months (95% CI, 5.4-9.9 months) and 7.5 months (95% CI, 4.4-10.6 months), respectively, for C1 and C2 (P¼ .761). The mPFS was 2.2 months (95% CI, 1.9-2.6 months) and 3.8 months (95% CI, 2.6-3.5 months), respectively, for C1 and C2 (P ¼ .005). The Kaplan-Meier curves for PFS and OS are shown inSupplemental Figures 1and2(in the online version).

Figure 1 Kaplan-Meier Curve of Overall Survival (OS) for the 3 Prognostic Groups

Abbreviation: DiM_{¼ Di Maio.}

(6)

-Discussion

In the present analysis, we found that the DiM prognostic score, initially developed and validated for patients with aNSCLC receiving 2L CHT, also performed well for patients receiving ICIs, allowing for the identiﬁcation of patients with a good prognosis and a subgroup of patients with a very short life expectancy.

In recent years, IO has changed the survival of patients with aNSCLC, signiﬁcantly prolonging the mOS and offering an interesting possibility of obtaining a long-term beneﬁt for a few subjects. Nonetheless, only a small percentage of patients (18%-20%) will respond to ICIs as 2L therapy, with a mPFS of ~2 to 4 months.4-8

The identification of prognostic and/or predictive clinical factors and biomarkers remains a crucial topic. The early identification of those with and without a response to ICIs will be decisive in the attempt to avoid inadequate treatment and optimize the use of drugs in clinical practice, sparing unnecessary toxicity and costs. The identification will also be essential to detect those patients who might experience detrimental effects from hyper-PD.35-37Different clinical factors are currently under investigation, including ECOG PS, patient age, smoking status, hyponatremia, use of steroids and antibiotics, and the presence of liver, bone, and brain metastases. However, no consensus has been reached regarding the benefit of ICIs for these patients.38,39

In both studies by Di Maio et al29,30 of the development and validation of the DiM score, the patients were classified into 3 survival groups: best, intermediate, and worst prognosis categories. Theirfindings led to the identification of a subset of patients (worst category) with a poor prognosis. For the latter patients, the mOS

was< 4 months. Although the ﬁnding represents prognostic, and not predictive, information, the investigators concluded that the possibility of beneﬁting from active treatment was very small for this category of patients.29,30

To indirectly compare our cohort of patients treated with ICIs with the patients who had received 2L CHT, we performed an external validation of the score in the same setting. Therefore, we speculated that if the score was successfully validated in our series of patients who had been treated with ICIs, these results could conﬁrm its prognostic role.

The validation of the DiM score in the present 347 patients treated with ICIs also allowed us to perform an indirect comparison between patients who had received CHT and those who had been treated with ICIs.

The results from our study have demonstrated that the use of the DiM score allowed for patient distribution into 3 survival categories for the patients who had received 2L ICI therapy. The C-index for OS was good (0.670), indicating satisfactory discrimination using the 3 risk categories. Speciﬁcally, patients within the worst prog-nosis group had the shortest mOS and mPFS at 2.6 and 1.9 months, respectively. It is probable that these patients would respond poorly to any anticancer therapy and that best supportive care could be the best choice for them. The differences in mPFS between the 2 centers participating in the present analysis might be explained by the differences in the timing of the radiologic assess-ments, which was longer in C2 compared than in C1. Similar to the results from other reports,22,23,38,39 _{our results highlighted the}

negative prognostic role of ECOG PS 2 for patients with aNSCLC and treated with ICIs.

Figure 2 Kaplan-Meier Curves of Progression-free Survival (PFS) for the 3 Prognostic Groups

Abbreviation: DiM_{¼ Di Maio.}

(7)

-The treatment of patients with a poor ECOG PS with ICIs has remained an argument of clinical debate. However, patients with a poor ECOG PS have a poor prognosis and, thus, are less likely to beneﬁt from ICIs, probably owing to their ineffective immune system and fewer functional lymphocytes. However, ongoing pro-spective trials are assessing the efﬁcacy of ICIs (ClinicalTrials.gov

identifiers, NCT02733159 and NCT02879617) in patients with a poor ECOG PS, which should help to better define the role of ICIs in this setting.40,41Because these drugs have been characterized by a favorable toxicity profile, many clinicians could be tempted to consider many patients who would have been excluded from CHT as being eligible for ICI therapy. Thus, after approval of ICIs in clinical practice in the setting of advanced urothelial cancer, the initiation of ICIs near the end of life significantly increased for patients with a poor PS although their use did not significantly change for patients with good PS.42 Again, in our study, the multivariate analysis results confirmed ECOG PS as a prognostic factor, suggesting that this important factor drives OS.

Patient sex is 1 of the factors included in the DiM prognostic score. Traditionally, in the CHT era, several analyses showed that men will have slightly worse results than those of women.43

Nevertheless, a recently reported meta-analysis of trials testing ICIs reported a significant interaction between ICI efficacy and sex, with worse outcome for women, probably resulting from a high occurrence of driver mutations.44However, another recent meta-analysis that included 23 randomized trials of ICI-treated patients did not reported differences when stratified by sex.45

Usually, patients with squamous cell NSCLC and especially those with rare histotypes (large cell neuroendocrine carcinoma, mixed and undifferentiated carcinoma) will have a poor prognosis compared with those with adenocarcinoma.46 Despite its negative prognostic role, patients with the squamous histologic type seemed to beneﬁt highly from ICI treatment,4,5 _{as did patients with}

adenocarcinoma. However, for those with rare histologic subtypes, the role of ICI treatment remains unclear.47

In the study by Di Maio et al of patients treated several years previously,30 patients who had received first-line platinum-based therapy had worse outcomes with 2L therapy. However, it is difficult to apply these findings to patients treated with ICIs, considering that, currently, most patients will receive first-line platinum-based CHT.37 However, the response to previous

therapies seemed to correlate with the response to ICIs in a retro-spective analysis.26

Ideally, the identification of predictive factors could improve decision making in clinical practice. With all the limitations of our indirect comparison, the improving of OS with ICI treatment for patients with aNSCLC was found when we compared the mOS for those in the best and intermediate categories in our present series compared with the patients who had received CHT in the original development of the DiM score (18.0 vs. 12.9 months and 8.5 vs. 6.9 months for the best and intermediate categories, respectively). Thus, when we compared the worst category results (4.0 vs. 2.6 months), the ICIs performed worse in terms of OS compared with CHT, possibly owing to the detrimental effect of ICIs in this group of patients. However, the patients included in the present analysis were treated in clinical practice, which probably allowed for the inclusion of some patients, who, because of poor PS, were excluded from the clinical trials of docetaxel used for the development of the DiM score. This could partially explain the worse outcome of the worst category in the present series. Consequently, our indirect comparison did not allow for a robust definition of the absolute benefit associated with ICIs in the different prognostic groups. Furthermore, it is important to emphasize that the score remains prognostic, rather than predictive. Finally, despite the poor out-comes, we cannot exclude that ICIs are also associated with activity and efficacy in the group of patients with the worst prognosis. However, the absolute outcomes in that group have undoubtedly been poor, and an honest and serious reflection is required on the cost-effectiveness of treatment with ICIs for these patients.

In addition to clinical factors,48_{multiple inﬂammatory markers}

have recently been investigated as possible prognostic and predictive biomarkers owing to their easy accessibility and limited testing costs.

Table 2 Description of DiM Scoring System29,30

Variable

Points

0 1 2 7

Sex Female Male

ECOG PS 0 1 2

Tumor stage III IV

Histologic type

Adenocarcinoma Squamous Other First-line therapy type Nonplatinum based Platinum-based ORR to ﬁrst-line Yes No

Abbreviations: DiM¼ Di Maio; ECOG ¼ Eastern Cooperative Oncology Group; ORR ¼ objective response rate; PS¼ performance status.

Table 3 Multivariate Analysis for OS for Overall Population

Stratiﬁed by Baseline Patient Characteristics

Variable HR 95% CI P Value Age 70 y 0.867 0.652-1.153 .327 Gender 1.003 0.749-1.344 .983 Smoker 1.303 0.904-1.880 .156 Pack-years₄₀ 0.712 0.528-0.960 .026 ECOG PS .000 0 vs. 2 0.140 0.092-0.213 .000 1 vs. 2 0.329 0.237-0.456 .000 Stage 0.895 0.492-1.630 .717 Histologic type 0.890 0.708-1.119 .320 Liver metastasis 1.729 1.268-2.358 .001 Brain metastasis 1.149 0.825-1.599 .412 Bone metastasis 1.226 0.935-1.607 .141 Second vs. third line or more 1.261 0.958-1.660 .099 Platinum-based ﬁrst-line therapy 1.083 0.659-1.780 .752 ORR atﬁrst-line therapy 0.818 0.628-1.065 .135

Abbreviations: CI_{¼ conﬁdence interval; ECOG ¼ Eastern Cooperative Oncology Group; HR ¼} hazard ratio; ORR_{¼ overall response rate.}

(8)

-The role of peripheral immune cells, through routine blood pa-rameters, was recently studied in patients treated with ICIs.49,50The neutrophil/lymphocyte ratio (NLR) has been the most studied because it better reﬂects the balance between protumor and anti-tumor activity of the host immune system.51,52Both Jiang et al53 and Cao et al54 reported that a greater baseline and post-treatment NLR was associated with poor PFS and OS.

The assessment of different biomarkers in a single predictive or prognostic score could allow for the identiﬁcation of patients who will have a survival beneﬁt from ICIs. Many immune-based scores were studied using clinical characteristics and blood biomarkers, such as iSEND (immunotherapy, sex, ECOG, NLR, delta NLR),11

ALI (advanced lung cancer inﬂammation index),18_{LIPI (lung}

im-mune prognostic index),24SII (systemic inﬂammation index),27and AISI (aggregate index of systemic inﬂammation).32_{All these scores}

included the NLR and most had included lactate dehydrogenase and the ECOG PS.55-58 Similar to our study, they identified different predictive or prognostic groups that were significantly associated statistically with a progressively worse PFS or OS. As confirmed in our multivariate analysis, the presence of baseline liver metastases and heavy smoker status ( 40 pack-years) have a rele-vant association with a worse prognosis. In the present analysis, our aim was to validate the prognostic role of the DiM score for patients receiving IO. However, liver metastases and smoking status were not among the factors evaluated for the development of the DiM score.29 _{However, the prognostic ability in this setting could be}

improved by including additional clinical and biologic factors. This was recently demonstrated in another study from our group. A score termed EPSILoN was created and then validated, which included 5 different prognostic parameters (ie, NLR, lactate dehydrogenase, smoking status, ECOG PS, and liver metastases).57

A major limitation of our study, just as for other recent studies that have tried to propose prognostic or predictive scores in patients treated with ICIs, was its retrospective nature, without a control group of patients who had not received ICIs. A control arm is necessary to assess the real predictive role of a marker or score. Moreover, the present study lacked the PD-L1 status of the patients included in the present analysis because, for some patients receiving ICIs within the expanded access program, PD-L1 tests were not required, especially in the beginning. Thus, its correlation with clinical factors was impossible. Moreover, when we compared the population between the 2 centers, some statistically signiﬁcant dif-ferences were found. For example, in C1, more women had been included compared with C2. This was probably because the smoking-related incidence of lung cancer in women increased in northern Italy (C1) compared with that in southern Italy (C2). This could also be the reason for treating more patients with stage III in C2 than in C1. Another difference was found among the patients with ECOG PS 2, which was more often present. This was probably because of the more accurate selection of patients within clinical trials (more frequent in C1) compared with those included in the Expanded Access Program (more frequent in C2). Adenocarcinoma was more common in C1 than in C2, probably because more female patients were included in C1 with a younger median age (65 vs. 67 years). The latter could also be the reason patients presented with more central nervous system, liver, and bone metastases at baseline before treatment using ICIs.

Finally, to the best of our knowledge, the present study is theﬁrst to apply a score originally developed for patients receiving CHT to a series of patients receiving IO to allow for an indirect comparison between these 2 treatments in different prognostic groups. Given the simple use, this score could be readily integrated into routine clinical practice, helping clinicians in decision-making.

Conclusions

The DiM score, generated using data from patients treated with 2L CHT, is also able to predict the prognosis of patients treated with ICIs. We recognize that, given the absence of a control group, its value remains prognostic, rather than predictive, for the efﬁcacy of IO. However, these results showed that patients within the worst category (poor clinical factors) had a short absolute life expectancy and probably would not beneﬁt from any active systemic therapy, regardless of the treatment type. Reasonably, for this subset of pa-tients, best supportive care could be the best choice. Regardless, integrating a composite biomarker (clinical and laboratory factors with molecular and PD-L1 status) is necessary for these patients, because the use of ICIs could even have a detrimental effect.

Clinical Practice Points

The development of prognostic or predictive biomarkers of the response to IO continues to be a great interest.

The use of a combination of biomarkers, rather than that of a single biomarker alone, could be a useful tool in decision making.

DiM (a clinical prognostic score) was previously validated in patients who had received 2L CHT.

The use of the DiM score was able to divide patients treated with 2L IO into 3 prognostic groups (best, intermediate, worst). The patients in the worst prognostic group did not beneﬁt from

IO or CHT; thus, best supportive care might be the best choice for these patients.

Acknowledgments

We would like to thank our nurse assistant, Anna Maria Leone, for her support.

Disclosure

M.G. has personal financial interests with AstraZeneca, MSD International GmbH, Bristol-Myers Squibb, Boehringer Ingelheim Italia SpA, Celgene, Eli Lilly, Ignyta, Incyte, Inivata, MedImmune, Novartis, Pfizer, Roche, and Takeda; institutional financial interests with Eli Lilly, MSD, Pfizer (MISP), AstraZeneca, MSD Interna-tional GmbH, Bristol-Myers Squibb, Boehringer Ingelheim Italia SpA, Celgene, Ignyta, Incyte, Inivata, MedImmune, Novartis, Pfizer, Roche, Takeda, Tiziana, and Foundation Medicine; and received research funding from AIRC, AIFA, Italian Moh, and TRANSCAN. G.L.R. reports personal fees from Eli Lilly, Bristol-Myers Squibb, and AstraZeneca, outside the submitted work. C.P. declares personal fees from Bristol-Myers Squibb and MSD, outside the submitted work. M.C.G. declares personal fees from MSD, AstraZeneca, Eli Lilly, and Bristol-Myers Squibb, outside the submitted work. D.S. declares personal fees from AstraZeneca,

(9)

-Boehringer Ingelheim, and Bristol-Myers Squibb, outside the sub-mitted work. The remaining authors declare that they have no competing interests.

Supplemental Data

Supplementalﬁgures accompanying this article can be found in the online version athttps://doi.org/10.1016/j.cllc.2020.01.005.

References

1.Califano R, Kerr K, Morgan RD, et al. Immune checkpoint blockade: a new era for non-small cell lung cancer. Curr Oncol Rep 2016; 18:59.

2.Reck M, Rodríguez-Abreu D, Robinson AG, et al. Pembrolizumab versus chemotherapy for PD-L1-positive non-small-cell lung cancer. N Engl J Med 2016; 375:1823-33.

3.Proto C, Ferrara R, Signorelli D, et al. Choosing wiselyﬁrst line immunotherapy in non-small cell lung cancer (NSCLC): what to add and what to leave out. Cancer Treat Rev 2019; 75:39-51.

4.Brahmer J, Reckamp KL, Baas P, et al. Nivolumab versus docetaxel in advanced squamous-cell non-small-cell lung cancer. N Engl J Med 2015; 373:123-35. 5.Borghaei H, Paz-Ares L, Horn L, et al. Nivolumab versus docetaxel in advanced

nonsquamous non-small-cell lung cancer. N Engl J Med 2015; 373:1627-39. 6.Herbst RS, Baas P, Kim DW, et al. Pembrolizumab versus docetaxel for previously

treated, PD-L1-positive, advanced non-small-cell lung cancer (KEYNOTE-010): a randomised controlled trial. Lancet 2016; 387:1540-50.

7.Rittmeyer A, Barlesi F, Waterkamp D, et al. Atezolizumab versus docetaxel in patients with previously treated non-small cell lung cancer (OAK): a phase 3, open-label, multicentre randomised controlled trial. Lancet 2017; 389:255-65. 8.Garassino MC, Cho BC, Kim JH, et al, ATLANTIC Investigators. Durvalumab as

third-line or later treatment for advanced non-small-cell lung cancer (ATLANTIC): an open-label, single-arm, phase 2 study. Lancet Oncol 2018; 19:521-36. 9.Mukherji D, Jabbour MN, Sarouﬁm M, et al. Programmed death-ligand 1

expression in muscle-invasive bladder cancer cystectomy specimens and lymph node metastasis: a reliable treatment selection biomarker? Clin Genitourin Cancer 2016; 14:183e7.

10.Chae YK, Pan A, Davis AA, et al. Biomarkers for PD-1/PD-L1 blockade therapy in non-small-cell lung cancer: is PD-L1 expression a good marker for patient selec-tion? Clin Lung Cancer 2016; 17:350e61.

11.Park W, Kwon D, Saravia D, et al. Developing a predictive model for clinical outcomes of advanced non-small cell lung cancer patients treated with nivolumab. Clin Lung Cancer 2018; 19:280-8.

12.Tay R, Prelaj A, Califano R. Immune checkpoint blockade for advanced non-small cell lung cancer: challenging clinical scenarios. J Thorac Dis 2018; 10(suppl 13): S1494-502.

13.Fucà G, Galli G, Poggi M, et al. Modulation of peripheral blood immune cells by early use of steroids and its association with clinical outcomes in patients with metastatic non-small cell lung cancer treated with immune checkpoint inhibitors. ESMO Open 2019; 4:e000457.

14.Fucà G, Galli G, Poggi M, et al. Low baseline serum sodium concentration is associated with poor clinical outcomes in metastatic non-small cell lung cancer patients treated with immunotherapy. Target Oncol 2018; 13:795-800. 15.Giulia G, Tiziana T, Claudia P, et al. Association between

antibiotic-immunotherapy exposure ratio and outcome in metastatic non-small cell lung cancer.Lung Cancer 2019; 132:72-8.

16.Prelaj A, Tay R, Ferrara R, et al. Predictive biomarkers of response for immune checkpoint inhibitors in non-small-cell lung cancer. Eur J Cancer 2019; 106:144-59.

17.Templeton AJ, Ace O, McNamara MG, et al. Prognostic role of platelet to lymphocyte ratio in solid tumors: a systematic review and meta-analysis. Cancer Epidemiol Biomarkers Prev 2014; 23:1204-12.

18.Shiroyama T, Suzuki H, Tamiya M, et al. Pretreatment advanced lung cancer inﬂammation index (ALI) for predicting early progression in nivolumab-treated patients with advanced nonesmall cell lung cancer. Cancer Med 2018; 7:13-20.

19.Russo A, Franchina T, Ricciardi GRR, et al. Baseline neutrophilia, derived neutrophil-to-lymphocyte ratio (dNLR), platelet-to-lymphocyte ratio (PLR), and outcome in non-small cell lung cancer (NSCLC) treated with nivolumab or docetaxel. J Cell Physiol 2018; 233:6337-43.

20.Ferrucci PF, Ascierto PA, Pigozzo J, et al. Baseline neutrophils and derived neutrophil-to-lymphocyte ratio: prognostic relevance in metastatic melanoma pa-tients receiving ipilimumab. Ann Oncol 2016; 27:732-8.

21.Weide B, Martens A, Hassel JC, et al. Baseline biomarkers for outcome of mel-anoma patients treated with pembrolizumab. Clin Cancer Res 2016; 22:5487-96. 22.Bagley SJ, Kothari S, Aggarwal C, et al. Pretreatment neutrophil-to-lymphocyte ratio as a marker of outcomes in nivolumab-treated patients with advanced non-small-cell lung cancer. Lung Cancer 2017; 106:1-7.

23.Diem S, Schmid S, Krapf M, et al. Neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR) as prognostic markers in patients with non-small cell lung cancer (NSCLC) treated with nivolumab. Lung Cancer 2017; 111:176-81.

24.Mezquita L, Auclin E, Ferrara R, et al. Association of the lung immune prognostic index with immune checkpoint inhibitor outcomes in patients with advanced non-small cell lung cancer. JAMA Oncol 2018; 4:351-7.

25.Nakaya A, Kurata T, Yoshioka H, et al. Neutrophil-to-lymphocyte ratio as an early marker of outcomes in patients with advanced non-small-cell lung cancer treated with nivolumab. Int J Clin Oncol 2018; 23:634-40.

26.Garde-Noguera J, Martin-Martorell P, De Julián M, et al. Predictive and prog-nostic clinical and pathological factors of nivolumab efﬁcacy in non-small-cell lung cancer patients. Clin Transl Oncol 2018; 20:1072-9.

27.Suh KJ, Kim SH, Kim YJ, et al. Post-treatment neutrophil-to-lymphocyte ratio at week 6 is prognostic in patients with advanced non-small cell lung cancers treated with anti-PD-1 antibody. Cancer Immunol Immunother 2018; 67:459-70. 28.Tanizaki J, Haratani K, Hayashi H, et al. Peripheral blood biomarkers associated

with clinical outcome in nonesmall cell lung cancer patients treated with nivo-lumab. J Thorac Oncol 2018; 13:97-105.

29.Di Maio M, Lama N, Morabito A, et al. Clinical assessment of patients with advanced non-small-cell lung cancer eligible for second-line chemotherapy: a prognostic score from individual data of nine randomised trials. Eur J Cancer 2010; 46:735-44.

30.Di Maio M, Krzakowski M, Fougeray R, et al. Prognostic score for second-line chemotherapy of advanced non-small-cell lung cancer: external validation in a phase III trial comparing vinﬂunine with docetaxel. Lung Cancer 2012; 77:116-20.

31.Schemper M, Smith TL. A note on quantifying follow-up in studies of failure time. Control Clin Trials 1996; 17:343-6.

32.Royston P, Parmar MK, Sylvester R. Construction and validation of a prognostic model across several studies, with an application in superﬁcial bladder cancer. Stat Med 2004; 23:907-26.

33.Pencina MJ, D’Agostino RB. Overall C as a measure of discrimination in survival

analysis: model speciﬁc population value and conﬁdence interval estimation. Stat Med 2004; 23:2109-23.

34.Eisenhauer EA, Therasse P, Bogaerts J, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 2009; 45: 228-47.

35.Ferrara R, Mezquita L, Texier M, et al. Hyperprogressive disease in patients with advanced non-small cell lung cancer treated with PD-1/PD-L1 inhibitors or with single-agent chemotherapy. JAMA Oncol 2018; 4:1543-52.

36.Champiat S, Ferrara R, Massard C, et al. Hyperprogressive disease: recognizing a novel pattern to improve patient management. Nat Rev Clin Oncol 2018; 15: 748-62.

37.Lo Russo G, Moro M, Sommariva M, et al. Antibody-Fc/FcR interaction on macrophages as a mechanism for hyperprogressive disease in non-small cell lung cancer subsequent to PD-1/PD-L1 blockade. Clin Cancer Res 2019; 25: 989-99.

38.Putzu C, Cortinovis DL, Colonese F, et al. Blood cell count indexes as predictors of outcomes in advanced non-small-cell lung cancer patients treated with nivolu-mab. Cancer Immunol Immunother 2018; 67:1349-53.

39.Tamiya M, Tamiya A, Inoue T, et al. Metastatic site as a predictor of nivolumab efﬁcacy in patients with advanced non-small cell lung cancer: a retrospective multicenter trial. PLoS One 2018; 13:e0192227.

40.Middleton G, Brock K, Summers Y, et al. Pembrolizumab in performance status 2 patients with non-small cell lung cancer (NSCLC): results of the PePS2 trial. Ann Oncol 2018; 29(suppl 8):viii493-547.

41.Passaro A, Spitaleri G, Gyawali B, et al. Immunotherapy in non-small-cell lung cancer patients with performance status 2: clinical decision making with scant evidence. J Clin Oncol 2019; 37:1863-7.

42.Parikh RB, Galsky MD, Gyawali B, et al. Trends in checkpoint inhibitor therapy for advanced urothelial cell carcinoma at the end of life: insights from real-world practice. Oncologist 2019; 24:e397-9.

43.Radzikowska E, Głaz P, Roszkowski K. Lung cancer in women: age, smoking, histology, performance status, stage, initial treatment and survival: population-based study of 20 561 cases. Ann Oncol 2002; 13:1087-93.

44.Conforti F, Pala L, Bagnardi V, et al. Cancer immunotherapy efﬁcacy and patients' sex: a systematic review and meta-analysis. Lancet Oncol 2018; 19:737-46. 45.Wallis CJD, Butaney M, Satkunasivam R, et al. Association of patient sex with

efﬁcacy of immune checkpoint inhibitors and overall survival in advanced cancers: a systematic review and meta-analysis. JAMA Oncol 2019; 5:529-36.

46.Pirker R, Pereira JR, Szczesna A, et al. Prognostic factors in patients with advanced non-small cell lung cancer: data from the phase III FLEX study. Lung Cancer 2012; 77:376-82.

47.Signorelli D, Ferrara R, Proto C, et al. Immune-checkpoints inhibitors in metastatic non-small cell lung cancer with rare histology. J Clin Oncol 2019; 37(suppl):9106. 48.Schmid S, Diem S, Li Q, et al. Organ-speciﬁc response to nivolumab in patients

with non-small cell lung cancer (NSCLC). Cancer Immunol Immunother 2018; 67: 1825-32.

49.Prelaj A, Pizzutilo P, Montrone M, et al. Prognostic and predictive role of pe-ripheral blood biomarkers in NSCLC patients treated with checkpoint, a single-center experience. J Thorac Oncol 2018; 13(suppl):S926.

50.Boeri M, Milione M, Proto C, et al. Circulating miRNAs and PD-L1 tumor expression are associated with survival in advanced NSCLC pa-tients treated with immunotherapy: a prospective study. Clin Cancer Res 2019; 25:2166-73.

51.Zer A, Sung MR, Walia P, et al. Correlation of neutrophil to lymphocyte ratio and absolute neutrophil count with outcomes with PD-1 axis inhibitors in patients with advanced non-small-cell lung cancer. Clin Lung Cancer 2018; 19:42.

(10)

-52.Fukui T, Okuma Y, Nakahara Y, et al. Activity of nivolumab and utility of neutrophil-to-lymphocyte ratio as a predictive biomarker for advanced non-small-cell lung cancer: a prospective observational study. Clin Lung Cancer 2019; 20:208-14.e2. 53.Jiang T, Qiao M, Zhao C, et al. Pretreatment neutrophil-to-lymphocyte ratio is

associated with outcome of advanced-stage cancer patients treated with immuno-therapy: a meta-analysis. Cancer Immunol Immunother 2018; 67:713-27. 54.Cao D, Xu H, Xu X, et al. A reliable and feasible way to predict the beneﬁts of

nivolumab in patients with non-small cell lung cancer: a pooled analysis of 14 retrospective studies. Oncoimmunology 2018; 7:e150726.

55.Cona MS, Lecchi M, Cresta S, et al. Combination of baseline LDH, performance status and age as integrated algorithm to identify solid tumor patients with higher

probability of response to anti PD-1 and PD-L1 monoclonal antibodies. Cancers (Basel) 2019; 11:223.

56.Prelaj A, Sara SE, Pizzutilo P, et al. Predictive score using clinical and blood biomarkers in advanced non small cell lung cancer (aNSCLC) patients treated with immunotherapy. Ann Oncol 2018; 29(suppl 10):x2.

57.Prelaj A, Ferrara R, Rebuzzi SE, et al. EPSILoN: a prognostic score for immu-notherapy in advanced non-small-cell lung cancer: a validation cohort. Cancers (Basel) 2019; 11:E1954.

58.Signorelli D, Giannatempo P, Grazia G, et al. Patients selection for immuno-therapy in solid tumors: overcome the naïve vision of a single biomarker. BioMed Res Int 2019; 2019:9056417.

(11)

-Supplemental Data

Supplemental Figure 1 Kaplan-Meier Curve of Progression-free Survival (PFS) for Center 1 and Center 2

(12)

-Supplemental Figure 2 Kaplan-Meier Curves of Overall Survival (OS) for Center 1 and Center 2