HER2 Expression May Hold the Key to Predicting Survival Outcomes in CRC

Heinz-Josef Lenz, MD, FACP

HER2 gene expression demonstrated value as a prognostic and predictive factor for treatment outcomes in patients with KRAS wild-type metastatic colorectal cancer (mCRC) enrolled in the phase 3 CALGB/SWOG-80405 trial (NCT00265850), with findings suggesting that this marker could help guide the selection of bevacizumab (Avastin)- vs cetuximab (Erbitux)-based regimens in this population, according to Heinz-Josef Lenz, MD, FACP.

In the study, next-generation sequencing (NGS) was used to assess HER2 amplification in primary tumor DNA from 559 patients, and NanoString gene expression was evaluated in tumor tissue from 925 patients using an 800-gene panel. Results showed that independent of treatment, high HER2 expression was linked with prolonged progression-free survival (PFS) and overall survival (OS). The median PFS was 11.6 months vs 10 months (P = .012) in the high- vs low-expression groups, and the median OS was 32 months vs 25.3 months (P = .033). In the high-expression group, cetuximab plus chemotherapy was associated with improved OS (P = .02) and a worse PFS for this combination was observed in the low-expression group (P = .019).

Moreover, when modeled as a continuous variable, higher HER2 expression was linked with longer OS (HR, 0.83; 95% CI, 0.75-0.93; adjusted P = .0007) and PFS (HR, 0.82; 95% CI, 0.74-0.91; adjusted P = .0002), which plateaued after the median. In those with HER2 expression below the median, treatment with the cetuximab combination led to inferior PFS outcomes vs the bevacizumab combination (HR, 1.38; 95% CI, 1.12-1.71; adjusted P = .0027); the same was true for OS (HR, 1.28; 95% CI, 1.02-1.59; adjusted P = .03).

“The most important message is that HER2 expression is not a mechanism of resistance to cetuximab. In fact, patients [being treated with cetuximab] seem to do better if they have high HER2 expression, so don’t use [that initial hypothesis] as an argument not to use [cetuximab],” said Lenz, who is a professor in the Department of Medicine and Department of Preventive Medicine, Division of Oncology, at the Keck School of Medicine in Los Angeles, California. “[Overall], I’m happy that we could shed some new light on [the role of] HER2 [in mCRC].”

In an interview with OncLive^®, Lenz expanded on the rationale for conducting this study, key findings regarding the correlation between HER2 expression and treatment outcomes in CRC and emphasized the potential utility of RNA sequencing to better understand the role of HER2 expression in CRC.

Lenz also serves as the associate director of Clinical Research, chair of the Gastrointestinal (GI) Oncology Program, and co-director of the Colorectal Center at the University of Southern California (USC) Norris Comprehensive Cancer Center in Los Angeles.

OncLive: What gaps in knowledge were you seeking to address by investigating the association between HER2 expression and outcomes with bevacizumab- vs cetuximab-based regimens in mCRC?

Lenz: We have learned over the past couple of years from other cancers like breast cancer and gastric cancer, that HER2 expression may be an important target for drug development. With the recent approvals of tucatinib [Tukysa] and fam-trastuzumab deruxtecan-nxki [Enhertu], it has become even more important to better understand how HER2 expression [plays a role] in metastatic disease. There are not a lot of studies looking at HER2 gene expression [in mCRC]. There has been a bit of [research using] immunohistochemistry [IHC], but we know from our colleagues in gastric and breast cancer that the classification of high and low HER2 expression using IHC can be tricky. [Therefore], we used RNA [testing] in a large, randomized phase 3 trial to see whether [HER2 expression] is predictive and prognostic. There have been data suggesting that high HER2 expression is a mechanism of resistance to cetuximab, so we also wanted to confirm the predictive value of HER2 in this randomized study.

Please describe the design and key objectives of the study.

The [original] study was designed to test the efficacy of cetuximab vs bevacizumab with physician’s choice of chemotherapy—the majority choose FOLFOX—to see whether patients with wild-type RAS have better overall survival outcomes with EGFR inhibitors. During this study, it became clear that [patients with] KRAS-mutant disease had EGFR resistance, so the study was put on hold and KRAS testing was introduced. At the end of the day, survival data were calculated based on approximately 1,100 patients randomized to either cetuximab- or bevacizumab-based treatment.

This study did not show any differences in OS. However, this study was one of the few in the world with comprehensive blood tissue collection, and incredible characterization of the molecular makeup [of mCRC], including liquid biopsy, germline testing, NGS, RNA sequencing, and proteomics. [Accordingly], we have an incredible wealth of [data showing the] molecular characterization of patients involved in this clinical trial.

All patients with wild-type RAS were included when they were newly diagnosed. [Other inclusion] criteria for this patient population were adequate liver function and renal function, and adequate hematologic parameters. We had two datasets to look at. One was the NGS database, where we also got some data on HER2 amplification. [Tumor tissues from] 925 patients [was tested for] NanoString [gene expression] using quantitative PCR. We had a panel of 800 genes, which included HER2.

What was the distribution of HER2 expression in this patient population?

First, we found that HER2 amplification is extremely rare, and the numbers are so small that statistical analysis would not make any sense. However, we included the amplification with a sophisticated statistical method that would be in consensus with HER2 expression. We also saw that the distribution for HER2 expression was not a normal distribution. Above the median, there was a plateau of the effect. We looked at the median, and that was calculated based on statistical methodology.

What did findings indicate about the prognostic value of HER2 expression levels for treatment outcomes in mCRC? Were any of these results particularly surprising?

The first finding was that HER2 expression was prognostic. If the HER2 level was high above the median, the median PFS increased from 10 to 11.6 months, and the OS increased from 25.3 to 32 months...Because this study was [conducted] in the overall patient population, the question was whether [HER2 expression] interacted with treatment [responses]. If HER2 expression was above the median, PFS and OS with cetuximab were significantly better. If the median was low, bevacizumab was better than cetuximab regarding PFS and [OS]. We always thought that [patients with] high levels [of HER2 expression] would be resistant to cetuximab, but that does not seem to be correct.

The data clearly showed that HER2 is highly prognostic for better PFS and OS and that there is a significant treatment effect. Below the median, bevacizumab is the treatment of choice; above the median, cetuximab is the treatment of choice. These are fairly new insights in the understanding of how to think about HER2.

What limitations from this study may be important to note?

We need to be very cautious [interpreting these data]. Mechanisms of resistance we identify in refractory patients in second and third line may not reflect the mechanism of resistance when you look at first-line treatments. There are a couple of examples in the literature, [such as] KRAS mutations [that we] think is a mechanism of resistance to EGFR inhibitors in the third and fourth line, but [not in the] first line.

[Notably, assessing] the expression level [using] RNA is not the standard of care, because it’s not a clear approved technology. Gene expression is typically performed using platforms that are not clearly approved, so you cannot use it in order to make treatment decisions in a clinical setting. There are a lot of RNA sequencing platforms that are approved, so [that’s] coming along very slowly. Also, NanoString technology is being used to potentially classify colon cancer. There are [delays] in directly implicating and implementing this finding in the clinic because of a lack of approved HER2 gene expression analysis. In the future, [however], I think this will hopefully help us to better select our patients for treatment options.

What next steps are planned for this research?

There is now significant investment in HER2-targeted drugs. These data may help to develop and successfully accelerate drug testing in this patient population. We also have started a collaboration with the National Cancer Center East in Tokyo to validate these HER2 findings in their randomized phase 3 PARADIGM trial [NCT02394795], which also tested bevacizumab vs panitumumab [Vectibix]. These data will be presented at this year’s ASCO Annual Meeting.

Overall, how might these findings support the utilization of RNA-based testing for HER2 expression when determining treatment strategies in CRC?

RNA can be a more quantitative measure for better understanding HER2 expression in tumors, [accounting for] the limitations of IHC. With this, I hope that we have a new understanding [of how] to use HER2 as a predictive and prognostic marker, but also open [a path] to integrate more RNA technology into molecular diagnostics.

Reference

Battaglin F, Ou F-S, Qu X, et al. HER2 gene expression levels are predictive and prognostic in patients with metastatic colorectal cancer enrolled in CALGB/SWOG 80405. J Clin Oncol. Published online March 8, 2024. doi:10.1200/JCO.23.01507