Tumor Growth Rate After Nadir Is Associated With Survival in Patients With EGFR-Mutant Non–Small-Cell Lung Cancer Treated With Epidermal Growth Factor Receptor Tyrosine Kinase Inhibitor

PURPOSE

To investigate the association between tumor volume growth rate after the nadir and survival in patients with EGFR-mutant advanced non–small-cell lung cancer (NSCLC) treated with erlotinib.

MATERIALS AND METHODS

Seventy-one patients with EGFR-mutant advanced NSCLC treated with erlotinib were studied for computed tomography tumor volume kinetics during therapy. The tumor growth rate after nadir was obtained using a previously published analytic module for longitudinal volume tracking to study its relationship with overall survival (OS).

RESULTS

The median tumor volume for the cohort was 19,842 mm³ at baseline and 4,083 mm³ at nadir. The median time to nadir was 6.2 months. The tumor growth rate after nadir for log_eV (the natural logarithm of tumor volume measured in mm³) was 0.11/mo on average for the cohort (SE: 0.014), which was very similar to the previously validated reference value of 0.12/mo to define slow and fast tumor growth. The OS of 48 patients with slow tumor growth (≤ 0.12/mo) was significantly longer compared with 23 patients with fast tumor growth (> 0.12/mo; median OS: 37.8 v 25.0 months; P = .0012). In Cox models, tumor growth rate was also associated with survival (regression coefficient: 3.9903; P = .0024; faster rate leads to increased hazards), after adjusting for time to nadir (regression coefficient: –0.0863; P = .0008; longer time to nadir leads to decreased hazards) and smoking history.

CONCLUSION

In patients with EGFR-mutant advanced NSCLC treated with erlotinib, slower tumor growth rates after nadir were associated with longer OS, providing a rationale for using tumor growth rates to guide precision therapy for lung cancer.

INTRODUCTION

Molecular targeted therapy has become the mainstream of the treatment of patients with advanced non–small-cell lung cancer (NSCLC) with specific genomic driver mutations.^1-4 In patients with advanced NSCLC harboring sensitizing mutations of epidermal growth factor receptor (EGFR) treated with EGFR tyrosine kinase inhibitors (EGFR-TKIs), the tumors typically show initial marked shrinkage, followed by further subsequent decrease leading to the nadir (the smallest tumor burden since the initiation of therapy), which is most commonly noted between 6 and 12 months of therapy.^5,6 After reaching the nadir, the tumors start to grow back and eventually progress, because of the development of different mechanisms of acquired resistance.^5-8 The tumors grow back and progress slowly over the course of a few years in some patients, whereas they may grow back more rapidly in other patients.^5,6,9-14

CONTEXT

• Key Objective

• In patients with epidermal growth factor receptor (EGFR)–mutant advanced non–small-cell lung cancer (NSCLC) treated with EGFR-tyrosine kinase inhibitor, tumor volume dynamics demonstrate a characteristic pattern with initial marked shrinkage, followed by subsequent eventual tumor regrowth because of acquired resistance. Tumors may grow back slowly over a few years in some patients, but may grow back more rapidly in other patients. The present study investigated the association between tumor growth rate after the volume nadir and survival in these patients.

• Knowledge Generated

• Slow tumor growth rates after nadir are associated with longer overall survival in patients with EGFR-mutant advanced NSCLC treated with erlotinib, which remained significant after adjusting for time to nadir and other clinical variables. To our knowledge, this is the first report that demonstrates the association between tumor growth rate and survival in these patients.

• Relevance

• This study provides a scientific basis for using tumor growth rate analyses to guide treatment decision making for molecular targeted therapy for advanced NSCLC.

Tumor volume dynamics of advanced NSCLC during effective targeted therapy can be reliably assessed using computed tomography (CT) tumor volumetry using clinical chest CT scans with high inter- and intraobserver agreements.^15,16 The pattern of initial response, nadir, and subsequent regrowth is reproducibly observed in patients with EGFR mutations and in patients with anaplastic lymphoma kinase (ALK) rearrangements treated with effective targeted therapy.^5-8,16-19 As an example, our previous report has characterized the volumetric tumor growth rate after nadir in patients with EGFR-mutant advanced NSCLC treated with erlotinib or gefitinib and provided a reference value of 0.12/mo for the logarithm of the volume (log_eV) as an overall tumor growth rate in these patients using a linear mixed-effects model. The overall growth rate after nadir of 0.12/mo was reproduced in the external validation cohort, demonstrating the consistency of the tumor growth rate after nadir among EGFR-mutant patients treated with EGFR-TKIs.^5,6,19 The results indicated that the reference value of 0.12/mo for log_eV may be able to reproducibly define slow tumor growth for EGFR-mutant patients on EGFR-TKI therapy that can be used to help guide treatment decisions beyond progression by Response Evaluation Criteria in Solid Tumors (RECIST). Clinicians may find it useful to have objective criteria rather than the subjective and less precise guidance that is currently used in targeted therapy trials where the treatment is continued beyond RECIST progression if the investigator believes that the patient is receiving clinical benefits.

The purpose of the present study is to investigate the association between tumor growth rate after the volume nadir and survival in EGFR-mutant advanced NSCLC treated with EGFR-TKI.

MATERIALS AND METHODS

Patients

The study included 71 patients with advanced NSCLC harboring sensitizing mutations of EGFR treated with EGFR-TKI, erlotinib, as their first EGFR-directed therapy between June 2006 and July 2017 was included in this study to allow at least 3 years of follow-up. All patients had at least one dominant measurable lung lesion (≥ 10 mm in the longest diameter) on baseline chest CT before initiation of therapy.^5,18 All patients also had initial tumor volume decrease after starting erlotinib and experienced subsequent tumor growth after reaching nadir (the smallest tumor volume after initiating therapy) noted on at least two postnadir CT scans while on erlotinib.^5,18 Following the institutional review board approval, chest CT scans were retrospectively reviewed and the demographics and clinical characteristics were collected from the medical records.

Longitudinal Tumor Volume Tracking Using Serial CT Scans

The tumor volume measurements of dominant lung lesions (one lesion per patient) on the baseline chest CT and on all follow-up CT scans during erlotinib therapy were performed by a thoracic radiologist (T.H.), using a previously validated technique on the volume analysis workstation (Vitrea, Vital Images, Minnetonka, MN), as published previously.^5-7,15 In patients with more than one measurable lung lesion, the largest lung lesion was selected as a dominant lesion on the basis of the longest diameter of the lesion, as in previous studies.^5-7,15,17

The workflow for tumor volume measurements has been previously published in patients with advanced NSCLC treated with EGFR inhibitors and ALK inhibitors.^5-7,15,18,20 In brief, the axial chest CT images were displayed with a real-time interactive navigation and a reader manually selects a small region of interest within a tumor by a mouse click, determining a seed point. The software automatically segments the lesion from the surrounding lungs and adjacent structures, using a three-dimensional seed-growing algorithm. The reader then visually assesses the automatically segmented tumor contours and, if needed, manually adjusts the contour to determine the final tumor contour. After segmentation and manual correction, the software automatically calculates tumor volume (measured in mm³).^5-7,15,18,20 The intra- and interobserver variability of the tumor volume measurements using this technique in patients with advanced NSCLC has been previously published, documenting a high reproducibility with an interobserver concordance correlation coefficient of 0.990.¹⁵

Calculation of Tumor Growth Rate Using the Analytic Module

The longitudinal tumor volume data for each patient including the baseline CT and all follow-up CT scans during erlotinib therapy were evaluated in the analytic module. The module is designed to automatically detect the tumor nadir and calculate the tumor growth rates after nadir along with a graphical display of tumor volume dynamics (Fig 1), as published before.¹⁹ The calculation of tumor growth rate used the natural logarithm (log_eV) of the tumor volume originally measured in mm,³ as described in previous studies.^5,19,21-24 The module computes the rate of change in log_eV by obtaining the line of best fit for all volume measurements from the nadir during therapy, using the least squares method.¹⁹ The slope of the best fit line, R, is automatically calculated by the module and represents the tumor growth rate (Fig 1).¹⁹

FIG 1.

Tumor volume tracking and tumor growth calculation by the analytic module. (A) A 63-year-old woman with EGFR-mutant adenocarcinoma of the lung treated with erlotinib, with the left upper lobe mass at baseline. (B) The tumor volume decreased during therapy and reached the nadir at 16.3 months. (C) After nadir, the tumor volume gradually increased over 1.5 years. (D) The trend graph for log_eV is displayed by the module on the Vitrea workstation and the tumor growth rate for log_eV calculated by the module was 0.024/mo. log_eV, logarithm of the volume.

Statistical Analysis

The overall survival (OS) was compared between the groups dichotomized at the growth rate of 0.12/mo for log_eV, using the previously validated reference value of tumor growth rate after nadir in EGFR-mutant patients treated with erlotinib.^5,6 The association between tumor growth rate after nadir and OS was further evaluated using Cox proportional hazards models, adjusting for time from the baseline to nadir and other clinical variables. The OS was measured from the initiation of therapy to death of any cause, and patients not experiencing the event by the time of analyses were censored at the last known date of follow-up. All P values are based on a two-sided hypothesis. A P value of < .05 was considered to be significant.

RESULTS

Table 1 summarizes demographics and clinical characteristics of 71 patients in the present study, as well as their tumor volume characteristics. All patients had advanced NSCLC at the time of starting erlotinib. The median age was 63 years, 54 patients (76%) were women, and 47 (66%) were never smokers. Erlotinib was given as the first-line systemic therapy in 67 patients and was given as the second-line therapy after prior cytotoxic chemotherapy in four patients. Twenty patients were still alive at the time of analyses, and 51 patients had died.

TABLE 1.

Demographics and Clinical Characteristics of 71 Patients

The median tumor volume for the cohort was 19,842 mm³ at baseline (range: 2,042-243,819 mm³) and 4,083 mm³ at nadir (range: 176-45,921 mm³), with the median time to nadir of 6.2 months (range: 1.4-37.8 months). The tumor growth rate after nadir for log_eV/mo was 0.11/mo on average for the cohort, with the SE of 0.014, which was very close to the reference value of 0.12/mo obtained in the prior cohorts of EGFR-mutant NSCLC treated with erlotinib or gefitinib using a linear mixed-effects model, as previously reported (Table 2).^5,6

TABLE 2.

Tumor Volume Characteristics of 71 Patients

On the basis of the results of the tumor growth rate in the present group and the reference value obtained and validated in previous studies, the patients were dichotomized at the reference growth rate of 0.12/mo for log_eV, to study the relationship between tumor growth rate and OS. Forty-eight patients had tumor growth rates ≤ 0.12/mo (slow growth), and 23 patients had tumor growth rates > 0.12/mo (fast growth). Median OS of 48 patients with slow growth was significantly longer than that of 23 patients with fast growth (Median OS: 37.8 v 25.0 months, respectively, log-rank P = .0012; Fig 2).

FIG 2.

Kaplan-Meier estimates of OS of patients dichotomized at 0.12/mo for log_eV for slow versus fast growth. Forty-eight patients with tumor growth rates ≤ 0.12/mo (slow growth) had significantly longer survival compared with 23 patients with tumor growth rates > 0.12/mo (fast growth; median OS: 37.8 v 25.0 months, respectively; log-rank P = .0012). log_eV, logarithm of the volume; OS, overall survival.

To further study the association between tumor growth rate and OS adjusting for time to nadir and other clinical variables, the Cox proportional hazards model was built to predict H(t), which is a cumulative hazard function at time t, measured in months from the baseline. Negative logarithmic transformation of cumulative survival function S(t) provides H(t) (H(t) = –log_eS(t)). h(t) is the quotient of death density function f(t) divided by survival function S(t) at time t, called hazard function (h(t) = f(t)/S(t)). h₀(t) is the unadjusted hazard function for patients with EGFR-mutant NSCLC receiving erlotinib without accounting for other variables, and S₀(t) represents unadjusted survival function. h(t) = h₀(t) × exp(Xβ), where h₀(t) and the log of hazard ratio (β) can be predicted independently. Building on the result, further Cox models were built to analyze the influence of tumor growth rate of volume after nadir on survival, accounting for time from baseline to nadir and other clinical variables. After assessment of the statistical significance and clinical relevance of these variables, the following model was obtained to predict h(t), which is the hazard ratio after adjusting for tumor growth rate (Gr), time to nadir (Nt), and smoking history (Sm),

$$\text{h}\left(\text{t}\right)={\text{h}}_{\text{0}}\left(\text{t}\right)\times \text{exp}\left(3.9903\times \text{Gr}-0.0863\times \text{Nt }+0.4583\times \text{Sm}\right)\text{,}$$

where Gr represents the growth rate after nadir (log_eV/mo), Nt represents the time to nadir from the baseline (measured in months), and Sm represents categorical smoking history and is 1 for former or current smokers and 0 for never smokers. The P values were .0024 for Gr, .0008 for Nt, and .1334 for Sm. The C-index for the model was 0.74.

With the regression coefficient of 3.9903 with the P value of .0024, the formula indicates that a more rapid growth rate is significantly associated with higher hazards for death. Similarly, with the negative regression coefficient of –0.0863 for Nt, longer time to nadir is significantly associated with decreased hazards for death. Positive smoking history of being either current or former smoker also contributes to increasing the risk of death with reference to never smoker, with the regression coefficient of 0.4583, although the P value did not reach the significant level. Smoking history was kept in the model not based on the statistical significance but based on the clinical relevance, because the model assesses the survival of patients with NSCLC and smoking is an important component of prognosis in these patients. Other variables including age, sex, race, or prior chemotherapy did not have significant impact on survival time (P > .12) and thus were not included in the final model. The number and types of subsequent therapies (P > .35) and EGFR mutation types (P = .456) were not significant predictors, either.

DISCUSSION

This study demonstrates that the slow tumor growth rates after nadir are associated with longer OS in patients with EGFR-mutant advanced NSCLC treated with erlotinib, which remains significant after adjusting for time to nadir and other clinical variables. To our knowledge, this is the first report that demonstrates the association between tumor growth rate after nadir and survival in these patients and provides a scientific basis for using tumor growth rate analyses to guide treatment decision making for molecular targeted therapy for advanced NSCLC.

The tumor growth rate after nadir of the present cohort was 0.11/mo on average for log_eV and was very similar to the reference value of 0.12/mo that was obtained in the prior cohorts of patients with EGFR-mutant NSCLC treated with erlotinib or gefitinib using a linear mixed-effects model.^5,6 The standard error was 0.014/mo, which was also very close to the previous study that showed the SE of 0.015/mo for the overall growth rate of 0.12/mo.⁵ Given these results, it was deemed reasonable to use 0.12/mo as the reference value to define slow versus fast growth in these patients, as previously proposed.^5,6 OS of patients with slow tumor growth (≤ 0.12/mo) was significantly longer than those who had fast tumor growth (> 0.12/mo; median OS: 37.8 v 25.0 months; P = .0012), which supports the use of the reference value to guide treatment decisions when tumors are regrowing after initial shrinkage. It should be noted that OS in the fast growth group was relatively long, with the median of 25.0 months, because these are patients who initially had tumor shrinkage, reached the nadir, and experienced tumor regrowth after nadir.

Since the time to nadir differs among patients, further survival analyses were performed for the association between the tumor growth rates and OS, adjusting for the time to nadir and other clinical variables using the Cox proportional hazards models. The final model showed that higher growth rate leads to increased hazards for death (regression coefficient: 3.9903; P = .0024) after adjusting for time to nadir and smoking history as relevant variables. The model also showed that longer time to nadir is significantly associated with decreased hazards for death (regression coefficient: –0.0863; P = .0008), as intuitively expected. Smoking history was also included as a covariate, as current or former smoking history increased the hazards for death (regression coefficient: 0.4583; P = .1334), although at the nonsignificant level. Other clinical variables are not significant in the model and thus not included in the final model. The results of the Cox model further confirmed that higher growth rates are associated with shorter survival and slower growth rates are associated with longer survival, independent of time to nadir and smoking.

The concept of tumor growth rate over time during therapy is not included in the RECIST guidelines. RECIST defined progressive disease when patients experience ≥ 20% (and 5 mm) increase of tumor size since the nadir, regardless of the length of time from the nadir.^26-28 However, clinically, patients with ≥ 20% increase in 2 months from the nadir versus ≥ 20% increase over 12 months may need to be managed differently. It has been suggested that important insights can be obtained by evaluating tumor growth rate during cancer therapy and supplement limitations of RECIST.^5,29-33 Previous studies evaluated tumor growth rate as a marker for defining meaningful trial end points and documenting drug efficacy in advanced solid tumors.^30,32-35 Although the potential role of tumor growth rates in enhancing detection of treatment effects is described in these studies, only a few studies evaluated the impact of tumor growth rates on the survival of patients. In clinical trials of renal cell carcinomas (RCC) and prostate cancers, the growth rate constant, obtained as log_e2/doubling time using tumor size, showed a negative correlation with survival,^32,33 indicating that faster tumor growth is associated with shorter survival in these tumors. In another study of patients with metastatic RCC treated in a phase III trial of multikinase inhibitor, sorafenib, tumor growth rate as an estimate of the increase of the tumor volume during the first cycle was associated with survival, both progression-free survival and OS, after adjustment to the standard prognostic score for metastatic RCC.³⁶

Tumor growth rate in patients with advanced lung cancer has also been recently recognized as an important marker of tumor response and progression in the setting of immune checkpoint inhibitor (ICI) therapy. A small subset of ICI-treated patients may experience accelerated tumor growth after starting ICI compared with the period before starting ICI, representing a phenomenon termed hyperprogressive disease.^8,37 Hyperprogressive disease in the setting of ICI was defined as RECIST progression at the first evaluation with ≥ twofold increase in the tumor growth rate during ICI therapy compared with the pretherapy period.³⁷ In patients with advanced NSCLC treated with ICI, hyperprogressive disease was noted in 13.8% and was associated with shorter survival.³⁸ However, the association between tumor growth rates and survival in patients with advanced NSCLC harboring oncogenic drivers treated with effective molecular targeting therapy, such as EGFR-mutant patients treated with EGFR-TKI, which is the focus of the present study, has not previously been reported. These patients tend to experience slow tumor growth over the course of many months to a few years after nadir, and thus, the tumor growth rate assessments in this context require longitudinal analyses using multiple serial scans, which benefit from dedicated software technology specifically developed for the purpose.¹⁹ The objective criteria supported by quantitative imaging technology will help to refine a rather subjective concept of continued clinical benefits as judged by providers, which has been used in studies and clinical trials of TKI continuation beyond RECIST progression.^39-43 Additionally, tumor growth rates can be correlated with other biomarkers for acquired resistance during targeted therapy, including cell-free DNA analysis that can detect resistance mutations from serial plasma sampling in patients treated with TKI.^44,45

The limitations of the study include a retrospective nature of the study design with patients treated at the single institution. Tumor volume tracking and growth rate assessments were performed using the volume measurement of a dominant lung lesion for each patient because we followed the approach used in the previous studies that demonstrated the prognostic value of tumor volume of one dominant lung lesion in patients with advanced NSCLC treated with EGFR-TKIs and ALK-TKIs.^5-7,15,17 The same approach was also used in the previous studies that defined the volumetric tumor growth rates in patients treated with EGFR-TKIs, which was validated in the external cohort, providing a reference value of tumor growth rate for this study.^5,6 Only one reader performed the measurements because high intra- and interobserver agreements have been previously published for the tumor volume measurements and the tumor growth rates.^15,19

All patients in the present study are treated with erlotinib, which is a conventional EGFR-TKI that has been used widely in patients with EGFR-mutant advanced NSCLC in the past decade. The study did not include newer third-generation agents such as osimertinib that was approved in 2018 as the first-line treatment in these patients. The data of patients treated with osimertinib in the clinical setting are currently accumulating, especially for tumor regrowth after nadir, which require follow-up over the course of a few years or longer. Further studies can be performed in osimertinib-treated EGFR-mutant patients once the sufficient follow-up data become available to investigate if the tumor growth rate after nadir is associated with OS in the setting of osimertinib. Additionally, the approach can be applied to patients with other genomic drivers such as ALK-rearranged patients treated with ALK inhibitors who demonstrate similar pattern of tumor volume dynamics with initial response and subsequent tumor growth after nadir.^17,18

In conclusion, slower tumor growth rates after nadir in patients with EGFR-mutant advanced NSCLC treated with erlotinib were associated with longer OS, providing a rationale for using tumor growth rates to guide precision therapy for lung cancer. The findings can be further validated in prospective cohorts of patients treated with EGFR-TKIs including newer agents such as osimertinib and can be extended to other cohorts of patients with different genomic drivers treated with effective targeting agents.

AUTHOR CONTRIBUTIONS

Conception and design: Mizuki Nishino, Hiroto Hatabu, Bruce E. Johnson

Administrative support: Bruce E. Johnson

Provision of study materials or patients: Junwei Lu, Pasi A. Jänne, Bruce E. Johnson

Collection and assembly of data: Mizuki Nishino, Takuya Hino, Natalie I. Vokes, Pasi A. Jänne

Data analysis and interpretation: Mizuki Nishino, Junwei Lu, Hiroto Hatabu, Bruce E. Johnson

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/po/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).