Human epidermal growth factor receptor 2 (HER2) extracellular domain levels are associated with progression-free survival in patients with HER2-positive metastatic breast cancer receiving lapatinib monotherapy

Authors


Abstract

BACKGROUND:

Changes in serum human epidermal growth factor receptor 2 (HER2) levels associated with clinical outcomes, including objective response rate, progression-free survival (PFS), and overall survival have been reported in patients with metastatic breast cancer (MBC) receiving trastuzumab and chemotherapy. This study investigated whether baseline or changes in serum HER2 correlated with overall response rate (ORR) and/or PFS in patients with MBC receiving first-line lapatinib monotherapy.

METHODS:

The EGF20009 study investigated lapatinib monotherapy in 138 HER2-positive patients with MBC previously untreated for their metastatic disease. Serum was collected and assessed at baseline and every 4 weeks for 16 weeks after treatment initiation. Disease assessment was performed at weeks 8 and 12 and every 12 weeks thereafter. A ≥20% decrease or increase in serum HER2 was defined as a significant change.

RESULTS:

Seventy-nine percent of patients had elevated baseline serum HER2. Baseline serum HER2 was associated with ORR (P = .043) but not PFS. Patients with a ≥20% decrease from baseline of serum HER2 at weeks 4, 8, 12, and 16 had a significantly increased ORR and prolonged PFS. Conversely, those with a ≥20% increase from baseline had a significantly lower ORR and shorter PFS.

CONCLUSION:

Significant decreases in serum HER2 levels during the first 16 weeks of lapatinib monotherapy were associated with better clinical outcome (longer PFS and increased ORR) in HER2-positive MBC patients. Cancer 2011;. © 2011 American Cancer Society.

The human epidermal growth factor receptor 2 (HER2, ErbB2) proto-oncogene is a transmembrane receptor with an extracellular domain (ECD) and intracellular tyrosine kinase activity.1, 2 HER2 is overexpressed or amplified in approximately 20%-25% of invasive primary breast cancers as measured by immunohistochemistry or fluorescence in situ hybridization (FISH), and it is associated with more aggressive disease.3, 4 The HER2 ECD is cleaved by a disintegrin and metalloprotease, and the remaining membrane-bound internal domain is constitutively activated.5 The ECD (p97-115 kDa) of the HER2 protein is released into the circulation, and serum HER2 levels are elevated in 30%-90% of patients with metastatic breast cancer (MBC).6, 7

An elevated pretreatment serum HER2 level is associated with decreased response to both first-line and second-line endocrine therapy.8, 9 Numerous studies have also shown that changes in serum HER2 parallel the clinical course of disease, with serum HER2 increasing with disease progression and decreasing with treatment response.10-13 In an international multicenter analysis of data from 307 MBC patients treated with first-line trastuzumab-based therapy, patients who achieved a >20% decrease in serum HER2 from baseline had a significantly higher overall response rate (ORR), longer time to progression (320 vs 180 days; P < .0001), longer duration of response (369 vs 230 days; P = .008), and longer overall survival (898 vs 593 days; P < .018) compared with patients who did not achieve a 20% decrease from baseline.14

Lapatinib (Tyverb, Glaxo Group Limited, Middlesex, United Kingdom; Tykerb in countries outside the European Union) is a selective small-molecule dual tyrosine kinase inhibitor that targets both human epidermal growth factor receptors 1 and 2 (HER1 and HER2). Unlike trastuzumab, lapatinib enters the cell and binds reversibly to the cytoplasmic adenosine triphosphate–binding site in the kinase domains of the epidermal growth factor receptor and HER2 receptors. This binding results in inhibition of receptor phosphorylation and subsequent inhibition of downstream signaling cascade through the ras/raf mitogen-activated protein kinase and phosphoinositol-3-kinase/Akt pathways, leading to decreased proliferation and increased apoptosis. This mechanism is distinct from that of trastuzumab, because synergistic antitumor activity of lapatinib with trastuzumab, as well as activity in trastuzumab-resistant cell lines, has been demonstrated.15 The combination of lapatinib and capecitabine significantly decreased the risk of disease progression relative to capecitabine alone in women with HER2-positive advanced or MBC previously treated with anthracyclines, taxanes, and trastuzumab.16 Moreover, single-agent lapatinib activity as first-line treatment in HER2-positive MBC was reported with an ORR of 24%.17 The purpose of this study was to determine whether changes in serum HER2 are associated with clinical outcomes in HER2-positive MBC patients receiving first-line lapatinib monotherapy.

MATERIALS AND METHODS

Patient Population

The EGF20009 study investigated lapatinib monotherapy in 138 HER2-positive patients with MBC who had not received trastuzumab previously.17 Amplification of HER2 was documented by way of FISH in primary or metastatic tumor tissues. Eligibility was determined via local laboratory assessment. In addition, all samples were centrally tested. Eligible patients were randomized to receive either 1500 mg lapatinib once daily or 500 mg lapatinib twice daily. Institutional review board approval was obtained, and each patient or their guardian signed informed consent. For the purpose of this biomarker analysis, patients who received either dose of lapatinib were pooled, because there was no significant difference in response between the 2 treatment groups. Where appropriate, these data conform to the REporting recommendations for tumor MARKer (REMARK) guidelines.18

Evaluation of Tumor Response and Clinical Endpoints

Disease state was assessed at weeks 8 and 12 and every 12 weeks thereafter using Response Evaluation Criteria in Solid Tumors (RECIST) and was reviewed by a blinded independent review committee. Progression-free survival (PFS) was defined as the interval from the time of randomization until the first documentation of disease progression or death due to any cause. Confirmed ORR was defined as the percentage of patients achieving either complete response or partial response.

Quantitation of the Serum HER2

Serum was collected every 4 weeks throughout the study. Serum HER2 levels were assessed at baseline and every 4 weeks for 16 weeks after initiation of lapatinib monotherapy or until disease progression. Serum HER2 was available from 138 patients at screening, 131 patients at week 4, 109 patients at week 8, 83 patients at week 12, and 61 patients at week 16. Serum HER2 was quantified using a commercially available test (Oncogene Science/WILEX, Inc., Cambridge, MA), which has been cleared by the US Food and Drug Administration (FDA) for follow-up and monitoring of patients with MBC as described previously.19 A significant decline of serum HER2 was defined as a decrease of ≥20% from baseline at the follow-up visit. Similarly, a significant increase of serum HER2 was defined as an increase of ≥20% from baseline at the follow-up visit. This definition of a significant change was derived using the FDA-cleared cutoff, which is 2.5 times the coefficient of variation of the HER2 assay.19 The 20% serum HER2 decline had also been reported as the optimal PFS cutoff obtained via receiver operating curve analysis of 302 patients treated with first-line trastuzumab-containing therapy.14 Baseline serum HER2 levels above 15 ng/mL were considered elevated.19

Statistical Analysis

The intent-to-treat population, comprising all women who underwent randomization and received at least 1 day of treatment, was used for the analyses. Results were correlated to PFS and ORR. This was an exploratory analysis with all time points weighted equally. There were no adjustments for multiple comparisons.

Baseline serum HER2 analysis

Patients were stratified into serum HER2–elevated or serum HER2–normal groups according to their week 0 serum HER2 values relative to the normal cutoff of 15 ng/mL. A log-rank test was used to compare PFS across serum HER2 strata. The Kruskal–Wallis test was used to compare the distribution of serum HER2 values for patients stratified by ORR.

Change in serum HER2 analysis

The distribution of change from baseline serum HER2 values for patients stratified by ORR and by partial response, stable disease, and progressive disease were compared using the Kruskal–Wallis test at each time point. Change from baseline was assessed by grouping patients who had ≥20% change (increase or decrease) from baseline serum HER2 levels and comparing them with groups of patients who did not achieve this change in serum HER2. For PFS, a log-rank test was used to compare change from baseline serum HER2 values across these strata. For ORR, the percentage of patients achieving response were compared using the Fisher exact test. Kaplan–Meier curves were produced for PFS and median PFS were calculated from these curves.

Multivariate analysis

PFS was analyzed using a multivariate Cox proportional hazards regression model to verify that the association of changes in patients' serum HER2 levels was not due to baseline patient characteristics. Covariates assessed for the Cox model were selected based on known clinical prognostic factors, and availability in the database. The following 10 covariates were included in the model: ≥20% change in serum HER2 at any time during weeks 4-16, visceral disease at baseline, number of sites of disease, stage of disease, bone metastasis, lung metastasis, liver metastasis, time from diagnosis of metastatic disease to randomization, hormone receptor status, and Eastern Cooperative Oncology Group performance status. Univariate proportional hazards models were used to screen the covariates, and those significant at P < .20 were included in a backward elimination multivariate Cox regression model according to a selection-stay criterion of P < .05. Baseline serum HER2 and baseline tumor lesion size were forced into the multivariate model.

Effects were considered significant if P < .05, and all P values were 2-sided. All analyses were conducted using SAS version 9.2.

RESULTS

EGF20009: Patient Characteristics and Clinical Activity

This was a retrospective analysis of serum samples from 138 patients with MBC who received first-line lapatinib monotherapy and who had not received prior trastuzumab treatment. All patients had HER2 gene amplification documented by FISH. The most common cause for lapatinib discontinuation was disease progression. In this study, 33 patients achieved partial response and 71 patients had stable disease on lapatinib monotherapy. Disease progression was reported in only 24 patients, and 10 patients were considered not evaluable.17

Baseline Serum HER2 and Clinical Outcomes

Of the 138 patients, 109 (79%) had a serum HER2 baseline above the normal cutoff of 15 ng/mL. Higher baseline serum HER2 levels were associated with ORR (P = .043) in patients receiving lapatinib (Figure 1). However, baseline serum HER2 levels were not associated with PFS (P = .87).

Figure 1.

Box plot of mean and median baseline serum human epidermal growth factor receptor 2 (HER2) levels for patients in the EGF20009 study is shown. Higher baseline serum HER2 levels were associated with overall response rate (P = .043) in patients receiving lapatinib.

Changes in Serum HER2 Differ by Clinical Outcome

Patients who achieved partial response while on lapatinib monotherapy experienced statistically significant decreases in median serum HER2 at weeks 4-16 compared with patients with stable disease or progressive disease (Figure 2). Patients with stable disease did not experience any significant changes in serum HER2, and patients whose disease progressed experienced an increase in median serum HER2 (Figure 3).

Figure 2.

Changes in median serum human epidermal growth factor receptor 2 (HER2) levels from screening through week 16 are shown. Patients achieving partial response while on lapatinib monotherapy experienced statistically significant decreases in serum HER2 from weeks 4 to 16. Patients achieving stable disease experienced no changes to median serum HER2, whereas patients experiencing progressive disease experienced increases in median serum HER2. The error bars represent the upper and lower quartiles of the serum HER2 change.

Figure 3.

Changes in median serum human epidermal growth factor receptor 2 (HER2) levels in patients responding to or not responding to lapatinib monotherapy from screening through week 16 are shown. Patients responding to lapatinib monotherapy experienced statistically significant decreases in median HER2 levels from weeks 4 to 16. In contrast, patients not responding to lapatinib did not experience any significant changes in median HER2 levels. The error bars represent the upper and lower quartiles of the serum HER2 change.

First Occurrence of ≥20% Decrease or ≥20% Increase from Baseline in Serum HER2

In a predefined analysis, serum HER2 from all 138 MBC patients was grouped according to the first week of significant decrease (≥20%) or the first week of significant increase (≥20%) from baseline at weeks 4, 8, 12, and 16 (Table 1). In 17 patients (12%), there was no significant increase or decrease detected in serum HER2 levels during the first 16 weeks. However, 64 (46%) of the 138 patients achieved a significant decrease of ≥20% by week 12. Of these 64 patients, 53 (83%) had a significant decrease by week 4, and none of the patients had a significant decrease in serum HER2 after week 12.

Table 1. First Occurrence of ≥20% Decrease or ≥20% Increase From Baseline Level of Serum HER2 in Patients Treated With Lapatinib Monotherapy
 Week 4Week 8Week 12Week 16Totala
  • a

    Total does not include patients who did not achieve ≥20% decrease or increase in serum HER2 levels (n=17).

  • b

    Seventy-four patients did not achieve ≥20% decrease of serum HER2 levels during the study.

  • c

    Four patients experienced significant increase of the serum HER2 levels between weeks 8 and 16.

  • d

    Seventy-four patients did not achieve ≥20% increase of serum HER2 levels during the study.

  • e

    Three patients experienced significant decrease of the serum HER2 levels between weeks 4 and 8.

No. of patients with decreaseb5383064c
No. of patients with increased42163364e

In contrast, 64 (46%) of the 138 patients had a significant increase in serum HER2 by week 16. Forty-two (66%) of these 64 patients had a significant increase in serum HER2 by week 4. By week 12, 61 (95%) of the 64 patients had a significant cumulative increase in serum HER2 of ≥20%. For the majority of patients, once they achieved a ≥20% decrease in their serum HER2, the level did not increase within the 16 weeks, illustrating that the change in serum HER2 is stable throughout the 16 weeks of treatment. Only 6.3% (4/64) of patients experienced a serum HER2 increase following an initial decrease, whereas only 4.7% (3/64) who had an initial increase had a subsequent decline over the first 16 weeks of treatment.

Association of Response Rate by 20% Decline or Increase in Serum HER2

The change in serum HER2 from baseline and its association with ORR was then evaluated. For patients achieving a ≥20% decline in serum HER2 (baseline to week 4), the ORR was significantly higher (24/52 = 46%) compared with those who did not have this decrease (9/73 = 12.3%) (P < .0001). This effect of significant association between a ≥20% decline in serum HER2 and ORR was also observed at weeks 8, 12, and 16 (data not shown).

Conversely, for patients who experienced a ≥20% increase in serum HER2 (baseline to week 4), the ORR was significantly lower (1/39 = 2.6%) compared with those who did not have this increase (32/86 = 37%) (P < .0001). This effect of significant association between a ≥20% increase in serum HER2 and ORR was also observed at weeks 8, 12, and 16 (Table 2).

Table 2. Response Separation by ≥20% Increase of Serum HER2 From Baseline
IncreaseResponse by <20% Increase, n/N (%)Response by ≥20% Increase, n/N (%)SeparationFisher Exact P
Week 432/86 (37)1/39 (3)34%<.0001
Week 830/68 (44)3/41 (7)37%<.0001
Week 1229/55 (53)1/28 (4)49%<.0001
Week 1627/42 (64)1/19 (5)59%<.0001
Increase at any time29/67 (43)4/61 (7)36%<.0001

Association of PFS by 20% Decline or Increase in Serum HER2

Patients who achieved a ≥20% decrease from baseline of serum HER2 at week 4 had a significantly longer median PFS of 28.4 weeks (95% confidence interval [CI], 20.7-36.1; n = 53) compared with 19.3 weeks (95% CI, 14.9-24.4; n = 78) for patients who did not experience the decrease (P = .016) (Figure 4). Similarly, patients achieving a ≥20% decrease at week 8 (Figure 5), week 12, and at any time by week 16 (Figure 6) showed a significantly longer median PFS than those patients who did not have the decrease.

Figure 4.

Progression-free survival is shown for patients who achieved a ≥20% decrease in serum human epidermal growth factor receptor 2 from baseline at week 4.

Figure 5.

Progression-free survival is shown for patients who achieved a ≥20% decrease in serum human epidermal growth factor receptor 2 from baseline at week 8.

Figure 6.

Progression-free survival is shown for patients who achieved a ≥20% decrease in serum human epidermal growth factor receptor 2 from baseline at any time by week 16.

Patients who experienced a ≥20% increase from baseline of serum HER2 at week 4 had a significantly shorter median PFS of 16.1 weeks (95% CI, 8.4-19.2; n = 42) compared with 28.4 weeks (95% CI, 20.2-36.1; n = 89) for patients who did not have the increase (P < .001). Similarly, patients experiencing a ≥20% increase at weeks 8 and 16 from baseline also had a significantly shorter PFS. Patients experiencing a ≥20% increase at week 8 from baseline had a significantly shorter PFS (median PFS of 16.2 weeks [95% CI, 11.7-20.1; n = 41]) compared with 28.4 weeks (95% CI, 20.7-36.2; n = 68) for those who did not experience the increase (P = .0006). Similarly, patients with a ≥20% increase at any time by week 16 also had a significantly shorter PFS (Figure 7).

Figure 7.

Progression-free survival is shown for patients who achieved a ≥20% increase in serum human epidermal growth factor receptor 2 from baseline at any time by week 16.

Multivariate Analysis: PFS

Univariate screening resulted in 7 covariates for consideration in the multivariate model for PFS: ≥20% change in serum HER2 at any time during weeks 4-16, visceral disease at baseline, number of sites of disease, lung metastasis, liver metastasis, time from diagnosis of metastatic disease to randomization, and Eastern Cooperative Oncology Group performance status. In addition, baseline serum HER2 and baseline tumor size were forced into the model. The univariate analysis of hormone receptor status had a P = .054. However, because hormone receptor data were incomplete (20% of patients with missing data), hormone receptor status was excluded from the multivariate analysis to maximize the number of patients included in the analysis.

The results revealed that only the change in serum HER2 (≥20% decrease in serum HER2 vs <20% decrease) and baseline tumor size were significant independent covariates for PFS (Table 3).

Table 3. Multivariate Model for Progression-Free Survival
VariableHazard Ratio (95% Confidence Interval)Pa
  • HER2 indicates human epidermal growth factor receptor 2; SLD, sum of longest diameters.

  • a

    P values are based on Wald chi-square with Breslow tie handling.

  • b

    Baseline tumor lesion size measured by SLD.

  • c

    Binary 1=Yes/0=No.

Baseline serum HER21.001 (0.999-1.002).2355
Baseline SLDb1.005 (1.000-1.010).0406
≥20% decrease from baseline in serum HER2 any time between weeks 4 and 16c0.290 (0.175-0.481)<.0001

In addition, an exploratory analysis was conducted using serum HER2 change as a continuous variable (eg, maximum percent change in ECD from baseline). The continuous ECD change results were consistent with the categorical ECD change results using the 20% cutoff: increases in serum ECD were associated with increased hazard of progression, and decreases in serum ECD were associated with decreased hazard of progression.

DISCUSSION

Baseline (pretreatment) serum HER2 as well as serial changes in serum HER2 up to 16 weeks postbaseline in 138 MBC patients treated with lapatinib monotherapy were investigated. All 138 women were HER2-positive by gene amplification FISH assay. Of the 138 patients in the study, 79% had elevated (>15 ng/mL) serum HER2 at baseline. Elevated serum HER2 at baseline (>15 ng/mL) was not associated with PFS while on lapatinib monotherapy, but was significantly correlated with ORR.

The serial change in serum HER2 from baseline was significantly associated with both patient response and PFS. A ≥20% decrease of serum HER2 from baseline to as early as 4 weeks was associated with prolonged PFS and higher ORR in patients receiving lapatinib monotherapy. This finding was confirmed at weeks 8, 12, and 16. Serum HER2 changes were generally consistent (ie, if the serum HER2 level declined, it generally remained decreased). However, due to limited serum HER2 data for time points beyond week 16, the stability of changes in serum HER2 cannot be assessed beyond the 16-week period. In general, the change in serum HER2 occurred early, because 74% of patients exhibited a significant change by week 4, and 93% by week 8.

This is the first study monitoring serum HER2 in MBC patients treated with lapatinib monotherapy and correlating serum HER2 with clinical outcome. Our results are consistent with previous reports measuring serum HER2 in trastuzumab-treated patients. Esteva et al20 showed that changes in serum HER2 levels from baseline predicted PFS in patients undergoing trastuzumab-based therapy for MBC; their report suggested that a rising serum HER2 may be detected sooner than clinical symptoms or radiologic progression. Kostler et al12 also reported that significant changes from baseline of serum HER2 predicted PFS, and that the change in serum HER2 preceded the clinical course of disease in trastuzumab-treated patients. In the largest study to date (307 patients with MBC), Ali et al14 reported that a 20% decrease of serum HER2 from baseline in the early weeks following therapy initiation predicted for significantly increased objective response rate, prolonged time to progression, and longer survival to first-line trastuzumab therapy. Similar results were obtained by other investigators.21

In conclusion, in this study baseline serum HER2 was not associated with PFS, but was with ORR. This result is similar to that of other recent large reports concluding that baseline serum HER2 did not predict response to trastuzumab22 or lapatinib therapy.23 However, in this study a 20% decrease from baseline in serum HER2 was associated with an increased response rate and prolonged PFS to lapatinib monotherapy. A significant decline (or increase) in serum HER2 levels as early as 4 weeks after the start of lapatinib therapy was correlated with response to therapy. Therefore, early changes in serum HER2 could be useful in monitoring clinical outcome in HER2-positive MBC patients receiving lapatinib monotherapy. For patients not achieving a 20% decrease from baseline in serum HER2, novel agents such as a disintegrin and metalloprotease sheddase inhibitor, which has been shown to enhance the antitumor effect of lapatinib in vitro,24 or other novel agents could be evaluated in clinical trials for the potential to improve patient response.25

Acknowledgements

We acknowledge the efforts of Henry L. Gomez, MD, who served as Principal Investigator of the EGF20009 study. We also thank Health Learning Systems for providing editorial assistance. Current address for Walter Carney: On-Q-ity, Waltham, Massachusetts. Current address for Julie Maltzman: Morphotek, Inc., Exton, Pennsylvania.

FUNDING SOURCES

Sponsored and funded by GlaxoSmithKline.

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

Ancillary