Sage Journals: Discover world-class research

Abstract

Background

This cohort study aimed to evaluate the impact of tumor burden (TB) on the efficacy of immunotherapy in patients with advanced non-small cell lung cancer (NSCLC).

Materials and Methods

Data from the POPLAR and OAK trials were extracted as the training and validation cohorts, respectively. TB was defined as the sum of the longest dimensions (blSLD) of measurable target lesions as per RECIST v1.1. The Kaplan-Meier curves and multivariate Cox regression analyses were performed to assess the association between TB with blood-tested tumor mutation burden (bTMB), PD-L1 expression, and survival outcomes. Additionally, random forest algorithms analysis was performed to evaluate the accuracy of TB in predicting 12-month mortality of NSCLC patients received atezolizumab.

Results

A total of 105 patients from the POPLAR trial and 322 patients from the OAK trial were recruited in the training and validation sets, respectively. Patients with TB-L have significantly better OS than those with TB-H in the training (mOS: 15.8 months vs 6.93 months) as well as the validation (mOS: 16.0 months vs 7.59 months) cohort. The multivariate Cox regression analysis indicated that TB is an independent biomarker for OS prediction, regardless of bTMB, PD-L1 expression, and number of metastasis sites. The impact of TB on 12-month mortality was expected to be stronger with the increase of TB, suggesting that patients with a high tumor burden experienced a detrimental effect on 12-month mortality.

Conclusion

TB may act as a prognostic biomarker for clinical benefit in NSCLC patients treated with immunotherapy alone. This may be potentially effective for predicting the efficacy of immunotherapy-based regimens.

Plain language summary

This study explored how the size of tumors influences the effectiveness of immunotherapy in patients with advanced non-small cell lung cancer (NSCLC). We analyzed data from two clinical trials involving NSCLC patients treated with the immunotherapy drug atezolizumab. Tumor burden (TB) was measured as the total size of all measurable tumors. The results showed that patients with smaller tumors had longer overall survival compared to those with larger tumors. This finding was consistent across both clinical trials. Even after considering other factors like tumor biomarkers (e.g., PD-L1 and blood-based tumor mutation burden) and the number of metastasis sites, tumor size remained an important predictor of survival. We also found that patients with larger tumors had a higher risk of dying within 12 months of treatment. These results suggest that tumor size can help predict which patients are more likely to benefit from immunotherapy. This information could guide doctors in tailoring treatment plans and selecting patients for immunotherapy-based treatments. In summary, measuring tumor size could provide valuable insights into how well lung cancer patients might respond to immunotherapy, making it an important tool for improving patient outcomes.

Keywords

non-small cell lung cancer immunotherapy PD-L1 inhibitor atezolizumab tumor burden biomarker

Introduction

Immune checkpoint inhibitors targeting programmed cell death-1 (PD-1) and its ligand PD-L1 have significantly transformed the therapeutic landscape of lung cancer, particularly non-small cell lung cancer (NSCLC).^1-3 Anti-PD-1/L1 therapies have been shown to confer long-term survival benefits in NSCLC patients, which has become a standard of care for driver-gene-negative NSCLC patients.^4,5 Despite the remarkable success of immunotherapy in clinical practice, the clinical benefits of ICIs remains unsatisfactory among unselected NSCLC patients; only a subset of patients would achieve a durable clinical response from such treatment, especially immunotherapy alone.^6,7 Accumulating evidence has demonstrated that a poor response to immunotherapy may be linked to low level of PD-L1 expression, low tumor mutation burden (TMB), etc.^8-11 However, these biomarkers appear insufficient for accurately predicting the efficacy of immunotherapy, as even NSCLC patients with high PD-L1 expression or high TMB could experience early disease progression during immunotherapy.^4,6,7

Tumor Burden (TB) has recently gained attention as a potential predictive biomarker for clinical response to ICIs. When exploring reasons for the failure of immunotherapy, a previous study found that it was not solely due to the inability of ICIs to stimulate an anti-tumor immune response, but also because of the imbalance between the strength of the anti-tumor immune response and baseline TB.¹² High TB may lead to insufficient reinvigoration of anti-tumor CD8 + T cells by ICIs in melanoma patients.¹² In a prior preclinical investigation, high TB was shown to contribute to a reduced capacity of tumor-infiltrating lymphocytes to effectively kill tumor cells, resulting in clinically ineffective immunotherapy.¹³ Indeed, these findings have been corroborated by recent clinical studies. Joseph RW, et al found that a high baseline TB was independently associated with an unfavorable survival outcome after pembrolizumab among melanoma patients in KEYNOTE-001.¹⁴ These findings raised a possibility that TB might serve as a predictive biomarker for the clinical benefit of immunotherapy in NSCLC. Subsequently, several previous small sample-size retrospective studies have also identify a negative association between TB and survival in NSCLC patients treated with PD-1 inhibitors.^15,16 However, it remains unclear whether this phenomenon also occurs in patients receiving PD-L1 inhibitors.

Thus, we hypothesized that TB-H would serve as an unfavorable prognostic biomarker for patients with advanced NSCLC treated with PD-L1 inhibitor, and TB could enhance the predictive accuracy of bTMB and PD-L1 expression. To test this hypothesis, the present study was performed using sequencing and clinical data from the OAK and POPLAR trials.

Materials and Methods

Patient Cohort

This cohort study is conducted according to STROBE guidelines using data from the randomized, open-label international Phase III, POPLAR (NCT01903993) and OAK (NCT02008227) trials, which have been described previously.^17-19 Patients with stage IIIB or IV NSCLC who had previously received chemotherapy were randomized to receive either atezolizumab or docetaxel. The data and materials from the OAK and POPLAR studies were derived from a prior publication.⁸ Atezolizumab groups were included in the analysis of how TB affects the survival of NSCLC patients receiving ICIs. Additionally, docetaxel groups were included to further analyze the influence of TB on the clinical benefit of atezolizumab vs docetaxel. The inclusion criteria were as follows: information on the baseline sum of the longest dimensions, as per Response Evaluation Criteria in Solid Tumor version 1.1 (RECIST v1.1) must be available; sequencing data must have passed the quality control, with a biomarker-evaluable population having sequence coverage of more than 800x and MSAF not less than 0.01. For the OAK trial, information on PD-L1 expression was also required. Institutional Review Board approval was not necessary, since the data used were de-identified, following the guidelines established by Gandara et al.⁸

Assessment of TB

TB was estimated clinically based on the baseline tumor size, which was defined by the sum of the longest dimensions (blSLD) of all measurable target lesions at baseline. blSLD for each patient was recorded in the trial data according to per RECIST v1.1. Time-dependent receiver operating characteristic (ROC) analysis and X-tile calculations were used to identify the optimal cutoff value of blSLD in both cohorts. This cutoff value was subsequently used to categorize patients into two groups: TB-high (TB-H) and TB-low (TB-L).

Outcomes

Tumor assessments were conducted at baseline and subsequently every six weeks. Investigator-assessments based on RECIST v1.1 were categorized as follows: complete response (CR), partial response (PR), stable disease (SD), and progressive disease (PD). The objective response rate was defined as the percentage of patients who achieved either CR or PR, while the disease-control rate was defined as the percentage of patients who achieved CR, PR, or SD. Progression-free survival (PFS) was defined as the duration from enrollment to PD or death due to any cause. Overall survival (OS) was defined as the duration from enrollment to death due to any cause.

Assessment of bTMB

Patients included in the POPLAR and OAK trials were required to provide archival blood samples available for exploratory analysis of immunotherapy-predictive biomarkers. As previously described, free circulating DNA was extracted from the pretreatment blood samples of each patient. The molecular profile was determined using the FoundationOne next-generation sequencing (NGS) assay, which analyzed a 1.1 Mb coding region of 315 cancer-related genes. The bTMB value was calculated by identifying base substitutions with an allele frequency of ≥0.5% and excluding germline mutations.⁸ The comprehensive analyses of the OAK and POPLAR trials, as well as the B-F1RST trial, demonstrated that the optimal cutoff for bTMB in predicting the clinical benefit of atezolizumab in NSCLC patients was 16 mutations/Mb.^8,10 Consequently, patients were categorized into TMB-H (TMB ≥16Mut/Mb) and TMB-L(TMB <16Mut/Mb) groups.

Assessment of PD-L1 Expression

PD-L1 expression was assessed in tumor samples using the VENTANA SP142 PD-L1 immunohistochemistry assay (Ventana Medical Systems, Inc., Tucson, AZ, USA). PD-L1 status was analyzed exclusively in patients from the OAK trials, as reported by Gandara et al.⁸ There are two classifications for PD-L1 expression levels: positive vs negative and high vs low. Positive PD-L1 expression (PD-L1, +) was defined as PD-L1 expression on 1% or more of tumor cells or tumor-infiltrating immune cells. Negative PD-L1 expression (PD-L1, -) was defined as PD-L1 expression on less than 1% of tumor cells or tumor-infiltrating immune cells. High PD-L1 expression (PD-L1, High) was defined as PD-L1 expression on 50% or more of tumor cells or 10% or more of tumor-infiltrating immune cells, while lower values are classified as low PD-L1 expression (PD-L1, Low).

Statistical Analysis

The demographics and clinicopathological characteristics of all patients were summarized using frequency and percentage for categorical variables, and median (range) for continuous variables. A time-dependent ROC curve was used to evaluate the accuracy of blSLD for predicting survival. The cutoff value for blSLD was determined using the highest Youden index (sensitivity + specificity - 1). Optimal cutoff points for blSLD were further validated using the X-tile software version 3.6.1. The Kaplan-Meier curve was used to estimate PFS and OS among different groups, while the Log-rank test and Gehan-Breslow-Wilcoxon test were applied to assess significant differences across survival curves. Given the nature of the subgroup analysis stratified by baseline characteristics and the potentially small sample size in subgroups, hazard ratios (HRs) with 95% confidence intervals (CIs) obtained from univariate Cox regression analysis were used to estimate the effect of TB on survival across various subgroups. Multivariate Cox regression analysis was conducted to determine the association between TB and survival. Random forest algorithms analysis was performed to predict the influence of each variable on 12-month mortality using all clinical features. All statistical analyses were performed using the MedCalc software (https://www.medcalc.org/, Version 20.0.3, Belgium), GraphPad Prism (https://www.graphpad.com, Version 8.3.0, USA), and R statistical software (version 4.1. https://www.R-project.org) with the packages “randomForest,” “ggplot2”, “pROC”, “ROCR”. A P-value of <0.05 was considered statistically significant.

Results

Patient Characteristics

According to the inclusion criteria, a total of 104 and 322 patients received atezolizumab in the POPLAR and OAK trials, respectively, were recruited for the training and validation sets (Supplemental Figure 1). The baseline characteristics of these patients are presented in Table 1. Patients included in the POPLAR cohort were younger and had a greater number of individuals who had received two prior regimens. No significant differences were observed in other baseline characteristics between the training and validation cohorts. Additionally, 106 and 215 patients receiving docetaxel in the POPLAR and OAK trails were also included.

Table 1.

Baseline Characteristics of Patients in the Training and Validation Cohorts.

Characteristic	POPLAR Cohort (n = 105)	OAK Cohort (n = 322)	P Value
Age
Median (range)	61.0 (42.0-82.0)	63.0 (39.0-82.0)	P = 0.03
Age ≥65 years	38 (36.2%)	147 (45.7%)	P = 0.11
Sex
Male	72 (68.6%)	202 (62.7%)	P = 0.29
Female	33 (31.4%)	120 (37.3)	P = 0.29
Race
White	78 (74.3%)	225 (69.9%)	P = 0.65
Asian	18 (17.1%)	68 (21.1%)	P = 0.65
Other	9 (8.6%)	29 (9.0%)
ECOG-PS
0	29 (27.6%)	114 (35.4%)	P = 0.15
1	76 (72.4%)	208 (64.6%)	P = 0.15
Histology
Squamous	38 (36.2%)	86 (26.7%)	P = 0.07
Non-squamous	67 (63.8%)	236 (73.3%)	P = 0.07
Prior therapy line
1	67 (63.8%)	246 (76.4%)	P = 0.02
2	38 (36.2%)	76 (23.6%)	P = 0.02
Tobacco use history
Never	17 (16.2%)	60 (18.6%)	P = 0.84
Current/previous	88 (83.8%)	262 (81.4)	P = 0.84
EGFR mutation
Positive	7 (11.7%)	29 (10.7%)	P = 0.99
Negative	53 (88.3)	242 (89.3%)	P = 0.99
Unknown	45	51
EML4-ALK fusion
Positive	0 (0%)	2 (1.2%)	P = 1.00
Negative	45(100%)	168 (98.8%)	P = 1.00
Unknown	60	152
KRAS fusion
Positive	9 (29.0%)	21 (21.6%)	P = 0.46
Negative	22 (71.0%)	76 (78.4%)	P = 0.46
Unknown	74	225
Metastases site	3 (1-8)	3 (1-8)	P = 0.34
blSLD(median, range)	68.6 mm (10-316)	80 mm (11-241)	P = 0.14
PD-L1 expression
TC3 or IC3	-	59 (18.3%)	-
TC0/1/2 or IC0/1/2	-	263 (81.7%)	-
TC0 and IC0	-	138 (42.9%)	-
TC1/2/3 or IC 1/2/3	-	184 (57.1%)	-

TB Analysis

The time-dependent ROC analysis, along with X-tile calculations, identified an optimal cutoff value of blSLD in the training cohort as 82 mm. Consequently, patients with blSLD ≥82 mm were classified as TB-H in the POPLAR and OAK cohorts. Correlation analysis revealed that TB-H was significantly associated with a greater number of metastasis sites (P < 0.05) in both the POPLAR and OAK cohorts. Additionally, a higher proportion of patients with ECOG PS = 1 was observed within the TB-H group in the POPLAR cohort (Supplemental Table 1). No other characteristics were significantly associated with TB.

Effect of TB on the Clinical Efficacy of Immunotherapy

In the POPLAR cohort, PFS was relatively longer in the TB-L group compared to the TB-H group; however, the difference was not statistically significant (median PFS:2.79 months vs 2.93 months; Figure 1A). Patients with TB-L exhibited significantly better OS than those with TB-H (median OS: 15.8 months vs 6.93 months). The 12- and 24-month OS rates were 56.7% and 38.7% for the TB-L group, and 43.6%, 21.4%, for the TB-H group, respectively (Figure 1B). In the validation cohort of the OAK study, both PFS and OS were significantly higher in the TB-L group than in the TB-H group (Figure 1(D) and (E)). The mPFS and mOS were 2.85 months and 16.0 months in the TB-L group, compared to and 2.53 months and 7.59 months in the TB-H group, respectively. The 3- and 6-month PFS rates were 48.4% and 33.1%, 38.5% and 26.5%, while the 12- and 24-month OS rates were 58.6% and 34.8%, 41.2% and 14.9%, in TB-L and TB-H groups, respectively. The ORR and DCR were relatively higher in the TB-L group than the TB-H group in both the training and validation cohort (Figure 1(C) and (F)), although these differences were not statistically significant. Forest plots were presented to illustrate the impact of TB on OS among NSCLC patients with various prespecified characteristics in both cohorts. As shown in Figure 2, the negative influence of TB-H on OS was evident in nearly all subgroups, particularly in the OAK cohort.

Figure 1.

The impact of TB on PFS (A, D), OS (B, E), and clinical response rate (C, F) in the training cohort (A-C) and validation cohort (D-F).

Figure 2.

Forest plot illustrating the association between tumor burden and OS stratified by prespecified characteristics among NSCLC patients receiving atezolizumab in both the training and validation cohorts.

Cox Analysis for TB in Predicting Survival of NSCLC Patients Receiving Atezolizumab

The results of the multivariate Cox regression analysis are presented in Table 2. In the POPLAR cohort, TB emerged as the independent prognostic factor for predicting OS after adjusting for all prespecified variables. Compared to patients with TB-H, those with TB-L exhibited a 42% reduction in the risk of death among all patients receiving atezolizumab (HR = 0.58, 95% CI 0.34-0.98). A similar finding was observed in the validation cohort. In the OAK cohort, the multivariate Cox regression analysis indicated that TB serves as an independent biomarker for OS prediction (HR = 0.55, 95% CI 0.42-0.74), independent with TMB, PD-L1 expression, the number of metastases site, and other factors (Table 2).

Table 2.

Multivariate COX Analysis for OS Prediction in the Training and Validation Cohorts.

Characteristic	POPLAR Cohort		OAK Cohort
Characteristic	HR with 95% CI	P Value	HR with 95% CI	P Value
Age
<65 yr vs ≥ 65 yr	0.85 (0.51-1.41)	0.53	1.08 (0.82-1.43)	0.57
Sex
Male vs female	0.83 (0.46-1.47)	0.52	1.19 (0.87-1.63)	0.27
Race
White vs Asian	2.04 (0.80-5.20)	0.14	1.52 (1.02-2.25)	0.04
Other vs Asian	1.78 (0.52-6.10)	0.36	1.99 (1.14-3.46)	0.01
ECOG-PS
1 vs 0	1.50 (0.82-2.74)	0.19	1.88 (1.40-2.55)	<0.01
Histology
Squamous vs non-squamous	1.32 (0.78-2.22)	0.30	1.48 (1.08-2.02)	0.02
Prior therapy line
2 vs 1	0.90 (0.53-1.53)	0.71	0.80 (0.57-1.11)	0.18
Tobacco use history
Never vs current	0.74 (0.25-2.19)	0.58	1.09 (0.61-1.94)	0.78
Previous vs current	1.28 (0.70-2.34)	0.43	1.19 (0.77-1.82)	0.43
TMB
TMB-L vs TMB-H	1.42 (0.75-2.70)	0.28	1.20 (0.86-1.69)	0.29
Metastases site
≥3 vs < 3	1.66 (0.92-3.03)	0.09	1.43 (1.07-1.90)	0.02
PD-L1 expression
High vs low	-	-	0.50 (0.33-0.74)	<0.01
Tumor burden
TB-L vs TB-H	0.58 (0.34-0.98)	0.04	0.55 (0.42-0.74)	<0.01

Influence of TB on the Predictive Ability of bTMB

Further, all patients in the training and validation cohorts were categorized into four groups (bTMB-H/TB-H, bTMB-H/TB-L, bTMB-L/TB-H, and bTMB-L/TB-L). In the POPLAR cohort, patients in the bTMB-L/TB-H group exhibited the shortest mPFS (1.45 months), with 3- and 6- month PFS rates of 38.2% and 20.6%, respectively, which are lower than those of the other groups (Figure 3A). Consistent with the PFS findings, bTMB-L/TB-H patients also demonstrated the poorest OS with a mOS of 6.90 months, significantly lower than that of the bTMB-L/TB-L group, which had an mOS of 15.8 months (Figure 3B). In the OAK cohort, survival analysis of PFS revealed that the bTMB-H/TB-L group had the best PFS at 4.21 months and OS at 15.7 months, while the bTMB-L/TB-H group had the poorest PFS at 1.94 months and OS at 6.67 months (Figure 3(D) and (E)). Regarding clinical response rates, the bTMB-H/TB-L group exhibited the highest ORR and DCR, whereas the bTMB-L/TB-H group had the lowest ORR and DCR in both cohorts (Figure 3(C) and (F)).

Figure 3.

The influence of TB on the predictive ability of bTMB. PFS (A, D), OS (B, E), and clinical response rate (C, F) are presented for both the training cohort (A-C) and the validation cohort (D-F).

Influence of TB on the Predictive Ability of PD-L1 Expression

Next, the patients in the OAK cohort were categorized into four groups based on PD-L1 expression: PD-L1(+)/TB-H, PD-L1(+)/TB-L, PD-L1(−)/TB-H, and PD-L1(−)/TB-L groups. PFS analysis did not reveal a significant difference among the four groups (Figure 4A). However, OS analysis indicated that both the PD-L1(+)/TB-L and PD-L1(−)/TB-L groups exhibited better mOS with the 12-month OS rates of 59.0% and 58.0%, respectively (Figure 4B). In contrast, the PD-L1(−)/TB-H group demonstrated the poorest mOS, with a 12-month rate of only 35.5%.

Figure 4.

The influence of TB on the predictive ability of PD-L1 expression in the OAK cohort (negative vs positive: (A-C), high vs low: (D-F)). PFS is shown in panels (A and D), OS in panels (B and E), and clinical response rate in panels (C and F).

Based on whether PD-L1 had high expression or not, the patients were divided into four groups: PD-L1(High)/TB-H, PD-L1(High)/TB-L, PD-L1(Low)/TB-H, and PD-L1(Low)/TB-L. Consistent results were observed in the PFS and OS analyses across these four groups, indicating that TB-H significantly shortened both PFS and OS in patients, regardless of whether they were in the PD-L1(High) or PD-L1(Low) category (Figure 4(D) and (E)). In line with the survival analysis, the ORR and DCR exhibited similar trends (Figure 4(C) and (F)).

Random Forest Algorithms to Predict 12-Month Mortality of NSCLC Patients Receiving Atezolizumab

A random forest algorithm analysis was conducted to predict 12-month mortality in NSCLC patients received atezolizumab, utilizing all clinical variables from the OAK trial. In terms of MeanDecreaseAccuracy and MeanDecreaseGini in the random forest algorithms, TB contributed more significantly to the prediction of 12-month mortality than bTMB, PD-L1 expression (Figure 5A). The influence of TB on 12-month mortality was expected to be stronger with an increase of TB, suggesting that patients with a high tumor burden experienced a detrimental effect on 12-month mortality (Figure 5B). The addition of TB into the predictive model resulted in an increase in its predictive ability (AUC = 0.623, 95%CI 0.565-0.681) when compared the model without TB (AUC = 0.569, 95% CI 0.500-0.638).

Figure 5.

Ranking of variable importance that associated with 12-month mortality using Random Forest (A); Larger MeanDecreaseAccurary and MeanDecreaseGini values indicate a stronger impact on 12-month mortality. (B) The partial effect of TB (mm) on 12-month mortality. (A) greater variable effect corresponds to a higher risk of death within 12-month.

Effect of TB on PFS, OS of NSCLC Patients Receiving Atezolizumab vs Docetaxel

As shown in Figure 6, atezolizumab may provide greater clinical benefits for NSCLC patients with low tumor burden, particularly in terms of OS. The OS of TB-L patients was improved with atezolizumab compared to docetaxel, with a hazard ratio of 0.62 (95% CI 0.49-0.97), which is more favorable than TB-H patients (HR = 0.76, 95% CI 0.59-0.97). Additionally, the difference in PFS between atezolizumab and docetaxel reached statistical significance in NSCLC patients with TMB-H/TB-L (HR = 0.61, 95%CI 0.37-0.98), PD-L1 (high)/TB-L (HR = 0.53, 95%CI 0.32-0.90). When stratified by TMB and PD-L1 expression, it remains evident that TB-L patients tend to experience longer PFS and OS from atezolizumab.

Figure 6.

PFS (A) and OS (B) of NSCLC patients receiving atezolizumab vs docetaxel.

Discussion

To our knowledge, this is the first study to evaluate the influence of TB on clinical outcomes in NSCLC patients treated with PD-L1 inhibitors, utilizing data from two prospective multicenter randomized controlled trials: POPLAR and OAK. TB was found to be strongly associated with the clinical outcomes of atezolizumab, especially for OS. TB may enhance the predictive accuracy of bTMB and PD-L1 in predicting survival among NSCLC patients receiving atezolizumab. Multivariate analysis indicated that TB serves as a significant prognostic biomarker for OS, independent of TMB, PD-L1, ECOG-PS, number of metastases site. Consequently, TB could potentially function as a stratification factor for the efficacy of atezolizumab in NSCLC patients with varying levels of TMB or PD-L1 expression.

These findings are rationally supported by previous clinical and experimental studies.^13,15,16,20 Several single-center pilot studies have observed an association between TB-H and worsened survival of ICIs in previously treated advanced NSCLC. Consistently, Suzuki S et al recently reported that a high pretreatment TB was associated with a poor outcome for PD-1 inhibitors in treatment-naïve NSCLC based on data from a retrospective cohort of 260 patients.²¹ Theoretically, three factors may contribute to the negative role of TB in immunotherapy. First, the previous investigations demonstrated that interstitial fluid pressure increased with tumor largening.²² Increased interstitial fluid pressure in the tumor led to the collapse of intra-tumoral lymphatic vessels resulting in tumor-associated antigen retention.²³ Consequently, the antigen-presenting cells are unable to migrate into the lymphatic system to activate the anti-tumor immune reactivity. Second, as the tumor grows, cancer cells could potentially develop mechanisms other than the PD-1/PD-L1 pathway to achieve immune escape. Expanding tumors produce numerous immune-suppressive cytokines and cells.^20,24,25 All these factors contribute to an immunosuppressive microenvironment in a large tumor, which may have induced a poor response to PD-1 or PD-L1 inhibitors. Third, Huang et al found that, in addition to failure of PD-1 to activate an anti-tumor immune response, the imbalance between the strength of anti-tumor immune response and pretreatment TB also plays a role.¹² Although T cells in TB-H patients could be reinvigorated by PD-1 blockade, it seems insufficient to attack the large tumor. Consistent results could be observed in analyzing effect of TB on survival of NSCLC patient receiving atezolizumab vs docetaxel, higher clinical benefit would be realized among TB-L patients.

Our study also identified an interesting phenomenon: the influence of TB was significantly greater among patients with TMB-L, PD-L1(−) or PD-L1(low), regardless of PFS or OS. This observation may be partly attributed to the imbalance between the immune response and the tumor. TMB-L or low PD-L1 expression reflected a weaker anti-tumor reactivity induced by ICIs, whereas TB-H indicated an excess of tumor cells that require robust reinvigoration of CD8 + T cells. This suggests that TB is a critical factor for predicting the efficacy of ICIs in NSCLC patients who lack beneficial biomarkers, such as low TMB and low PD-L1 expression. In addition, among patients with high PD-L1 expression or TMB-H, we observed that TB-H patients exhibited a poorer response to ICIs compared to those with TB-L, as indicated by ORR, DCR, PFS as well as OS. Although some differences were not statistically significant, this may explain why ICIs do not provide substantial clinical benefits for certain NSCLC patients harboring TMB-H or high PD-L1 expression. Recently, the combination of chemotherapy with ICI has shown superiority over chemotherapy alone. However, it remains unclear which treatment is more appropriate, for instance, whether pembrolizumab alone or pembrolizumab combined with chemotherapy is better for NSCLC with strong PD-L1 expression. These findings may suggest optimal treatment strategies for advanced NSCLC based on TB. Baseline TB may be a critical factor in determining whether to use ICIs alone or in combination with other treatments. Previous preclinical studies indicated that reducing tumor burden with conventional cytotoxic agents could overcome the resistance of large tumors to ICIs.¹³ Therefore, for large tumors, conventional chemotherapy or local therapy combined with ICIs may be appropriate, while ICIs alone should remain the first choice for TB-L patients with TMB-H or high PD-L1 expression. Given that the angiogenesis-related genes are highly expressed in patients with TB-H,²¹ anti-angiogenesis treatment may represent a potential alternative approach.

There are several advantages to using TB as a predictive biomarker for the efficacy of ICIs. First, TB is a clinically measurable biomarker that can be assessed in real-world clinical practice. Physicians can easily determine TB in almost any setting as long as they evaluate the pretreatment target regions by radiological examination. Second, obtaining TB is a non-invasive procedure that incurs no additional costs before initiating ICIs treatment. Our findings may also have implications for trial design in NSCLC. Given the strength of TB as an independent prognostic factor in both RCTs cohorts, TB could be considered a stratification factor for ICIs-related clinical trials. However, the application of TB to stratify patients can still be challenging. Firstly, there is currently no recognized gold standard for assessing TB. In our study, we utilized the blSLD per RECIST v1.1 as the indicator for TB. Some studies have used the number of metastatic sites,¹⁵ while others have modified RECIST v1.1 to include a maximum of 10 target lesions and five per organ.^14,21 Secondly, the cutoff point for TB has not been clearly defined. Different cohorts and various types of ICIs yield different cutoff points. We established 82 mm as the optimal cutoff point for blSLD in the training cohort when using time-dependent ROC and X-tile software. Interestingly, this cutoff point also proved to be the best cutoff point in the validation cohort, suggesting the homogeneity of these two cohorts.

The present study has several limitations. First, our method for estimating TB was based on the diameters of measurable lesions; however, we could not evaluate the TB directly. It remains uncertain whether the blSLD accurately represents TB. RECIST v1.1 is a framework for selecting measurable lesions to evaluate treatment response, but it may not fully capture the total disease burden. For example, non-target lesions such as pleural effusion, bone metastases, and other lesions that are difficult to measure may represent a significant additional tumor burden, yet these are not included in the sum of measurements. Additionally, it remains unclear about the influence of TB on efficacy among patients with the specific metastases sites, such as central-nerve system metastases, liver metastases, due to the inadequate data in these two trials. Second, the impact of TB on the predictive value of PD-L1 expression for immunotherapy was assessed solely in the OAK cohort, as data regarding PD-L1 status in the POPLAR cohort were unavailable. Nevertheless, our study utilized data from prospective multicenter trials, and our analysis, employing robust statistical methods on both the training and validation cohorts, consistently indicated a correlation between TB and the efficacy of atezolizumab. Further investigations are warranted to elucidate the mechanisms by which a high tumor burden contributes to the reduced efficacy of ICIs.

Conclusion

The present study evaluated the impact of tumor burden (TB) in NSCLC patients who were treated with atezolizumab. The results indicated that TB may serve as an independent prognostic biomarker for OS, regardless of TMB, PD-L1 expression, ECOG-PS, and the number of metastatic sites, and other factors. Furthermore, TB may function as a stratification factor for the efficacy of atezolizumab in NSCLC patients with varying levels of TMB or PD-L1. Additionally, TB could enhance the predictive accuracy of bTMB and PD-L1 in forecasting survival outcomes for NSCLC patients undergoing treatment with atezolizumab.

Supplemental Material

Supplemental Material - The Association Between Tumor Burden and the Efficacy of Immunotherapy Among Patients With Non-small Cell Lung Cancer

Supplemental Material for The Association Between Tumor Burden and the Efficacy of Immunotherapy Among Patients With Non-small Cell Lung Cancer by Jia-Jun Hui, Han-Lu Yan, Sheng-Jun Ding, Bao-Dong Qin, Xiao-Dong Jiao, and Yuan-Sheng Zang in Cancer Control.

Footnotes

Acknowledgements

We thank the patients for their participation in the OAK and POPLAR clinical trial, clinical trial, and the investigators for releasing the sequencing data and clinical data.

Author Contributions

Yuan-Sheng Zang designed the study. Jia-Jun Hui, Han-Lu Yan, Sheng-Jun Ding collected data. Jia-Jun Hui, Han-Lu Yan, Sheng-Jun Ding and Bao-Dong Qin did the analyses and prepared figures. Jia-Jun Hui, Han-Lu Yan, Sheng-Jun Ding and Xiao-Dong Jiao interpreted the data and wrote the manuscript. All authors gave final approval to submit for publication. All authors have agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All these authors listed read and approved the final manuscript.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Ethical Statement

ORCID iD

Bao-Dong Qin

Supplemental Material

Supplemental material for this article is available online.

Appendix

References

Paz-Ares

Luft

Vicente

, et al. Pembrolizumab plus chemotherapy for squamous non-small-cell lung cancer. N Engl J Med. 2018;379:2040-2051.

Mok

TSK

Kudaba

, et al. Pembrolizumab versus chemotherapy for previously untreated, PD-L1-expressing, locally advanced or metastatic non-small-cell lung cancer (KEYNOTE-042): a randomised, open-label, controlled, phase 3 trial. Lancet. 2019;393:1819-1830.

Garassino

Gadgeel

Esteban

, et al. Patient-reported outcomes following pembrolizumab or placebo plus pemetrexed and platinum in patients with previously untreated, metastatic, non-squamous non-small-cell lung cancer (KEYNOTE-189): a multicentre, double-blind, randomised, placebo-controlled, phase 3 trial. Lancet Oncol. 2020;21:387-397.

Reck

Rodriguez-Abreu

Robinson

, et al. Five-year outcomes with pembrolizumab versus chemotherapy for metastatic non-small-cell lung cancer with PD-L1 tumor proportion score ≥ 50. J Clin Oncol. 2021;39:2339-2349.

Borghaei

Ciuleanu

Lee

, et al. Long-term survival with first-line nivolumab plus ipilimumab in patients with advanced non-small cell lung cancer: a pooled analysis. Ann Oncol. 2023;34:173-185.

Carbone

Reck

Paz-Ares

, et al. First-line nivolumab in stage IV or recurrent non-small-cell lung cancer. N Engl J Med. 2017;376:2415-2426.

Herbst

Giaccone

de Marinis

, et al. Atezolizumab for first-line treatment of PD-L1-selected patients with NSCLC. N Engl J Med. 2020;383:1328-1339.

Gandara

Paul

Kowanetz

, et al. Blood-based tumor mutational burden as a predictor of clinical benefit in non-small-cell lung cancer patients treated with atezolizumab. Nat Med. 2018;24:1441-1448.

Stein

Rimm

, et al. Comparison of biomarker modalities for predicting response to PD-1/PD-L1 checkpoint blockade: a systematic review and meta-analysis. JAMA Oncol. 2019;5:1195-1204.

10.

Kim

Velcheti

Mekhail

, et al. Blood-based tumor mutational burden as a biomarker for atezolizumab in non-small cell lung cancer: the phase 2 B-F1RST trial. Nat Med. 2022;28:939-945.

11.

Peters

Dziadziuszko

Morabito

, et al. Atezolizumab versus chemotherapy in advanced or metastatic NSCLC with high blood-based tumor mutational burden: primary analysis of BFAST cohort C randomized phase 3 trial. Nat Med. 2022;28:1831-1839.

12.

Huang

Postow

Orlowski

, et al. T-cell invigoration to tumour burden ratio associated with anti-PD-1 response. Nature. 2017;545:60-65.

13.

Haratani

Yonesaka

Takamura

, et al. U3-1402 sensitizes HER3-expressing tumors to PD-1 blockade by immune activation. J Clin Investig. 2020;130:374-388.

14.

Joseph

Elassaiss-Schaap

Kefford

, et al. Baseline tumor size is an independent prognostic factor for overall survival in patients with melanoma treated with pembrolizumab. Clin Cancer Res. 2018;24:4960-4967.

15.

Miyawaki

Kenmotsu

Mori

, et al. Association between clinical tumor burden and efficacy of immune checkpoint inhibitor monotherapy for advanced non-small-cell lung cancer. Clin Lung Cancer. 2020;21:e405-e414.

16.

Sakata

Kawamura

Ichikado

, et al. Comparisons between tumor burden and other prognostic factors that influence survival of patients with non-small cell lung cancer treated with immune checkpoint inhibitors. Thorac Cancer. 2019;10:2259-2266.

17.

Rittmeyer

Barlesi

Waterkamp

, 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;389:255-265.

18.

Fehrenbacher

Spira

Ballinger

, et al. Atezolizumab versus docetaxel for patients with previously treated non-small-cell lung cancer (POPLAR): a multicentre, open-label, phase 2 randomised controlled trial. Lancet. 2016;387:1837-1846.

19.

von Elm

Altman

Egger

, et al. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann Intern Med. 2007;147:573-577.

20.

Katsurada

Nagano

Tachihara

, et al. Baseline tumor size as a predictive and prognostic factor of immune checkpoint inhibitor therapy for non-small cell lung cancer. Anticancer Res. 2019;39:815-825.

21.

Suzuki

Haratani

Hayashi

, et al. Association of tumour burden with the efficacy of programmed cell death-1/programmed cell death ligand-1 inhibitors for treatment-naive advanced non-small-cell lung cancer. Eur J Cancer. 2022;161:44-54.

22.

Raju

Haug

Ibrahim

Heyeraas

. High interstitial fluid pressure in rat tongue cancer is related to increased lymph vessel area, tumor size, invasiveness and decreased body weight. J Oral Pathol Med. 2008;37:137-144.

23.

Schaaf

Garg

Agostinis

. Defining the role of the tumor vasculature in antitumor immunity and immunotherapy. Cell Death Dis. 2018;9:115.

24.

Nagarsheth

Wicha

Zou

. Chemokines in the cancer microenvironment and their relevance in cancer immunotherapy. Nat Rev Immunol. 2017;17:559-572.

25.

Binnewies

Roberts

Kersten

, et al. Understanding the tumor immune microenvironment (TIME) for effective therapy. Nat Med. 2018;24:541-550.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.14 MB