Purpose:

To demonstrate the negative prognostic impact of a panel of genomic alterations (PRESSING-HER2 panel) and lack of HER2 amplification by next-generation sequencing (NGS) in patients with HER2+, RAS wild-type metastatic colorectal cancer receiving dual HER2 blockade.

Experimental Design:

The PRESSING-HER2 panel of HER2 mutations/rearrangements and RTK/MAPK mutations/amplifications was assessed by NGS. HER2 amplification was confirmed by NGS if copy-number variation (CNV) was ≥ 6. With a case–control design, hypothesizing 30% and 5% PRESSING-HER2 positivity in resistant [progression-free survival (PFS) <4 months and no RECIST response] versus sensitive cohorts, respectively, 35 patients were needed per group.

Results:

PRESSING-HER2 alterations included HER2 mutations/rearrangements, EGFR amplification, and BRAF mutations and had a prevalence of 27% (9/33) and 3% (1/35) in resistant versus sensitive patients (P = 0.005) and 63% predictive accuracy. Overall, HER2 nonamplified status by NGS had 10% prevalence. Median PFS and overall survival (OS) were worse in PRESSING-HER2+ versus negative (2.2 vs. 5.3 months, P < 0.001; 5.4 vs. 14.9 months, P = 0.001) and in HER2 nonamplified versus amplified (1.6 vs. 5.2 months, P < 0.001; 7.4 vs. 12.4 months, P = 0.157). These results were confirmed in multivariable analyses [PRESSING-HER2 positivity: PFS HR = 3.06, 95% confidence interval (CI), 1.40–6.69, P = 0.005; OS HR = 2.93, 95% CI, 1.32–6.48, P = 0.007]. Combining PRESSING-HER2 and HER2 CNV increased the predictive accuracy to 75%.

Conclusions:

PRESSING-HER2 panel and HER2 nonamplified status by NGS warrant validation as potential predictive markers in this setting.

See related commentary by Raghav et al., p. 260

Translational Relevance

The presence of a panel of candidate genomic resistance alterations (PRESSING-HER2 panel including HER2 mutations/rearrangements and mutations/amplifications in RTK/MAPK genes) and HER2 nonamplified status [HER2 copy-number variation (CNV) <6] assessed by means of NGS predicts primary resistance to dual HER2 blockade in HER2+ metastatic colorectal cancer. Thus, comprehensive genomic profiling may improve the negative selection for HER2 targeting, and the negative predictive value of PRESSING-HER2 and HER2 CNV warrants validation in ongoing randomized controlled trials. In patients with low predicted sensitivity to anti-HER2 targeted strategies, alternative options such as antibody–drug conjugates may bypass the genomic mechanisms of resistance. Finally, the use of NGS to select tumors without genomic codrivers of resistance and with confirmed HER2 amplification may identify patients with HER2 addiction who may benefit from chemo-free targeted strategies.

Several phase II nonrandomized trials have shown promising activity of dual HER2 blockade in pretreated patients with HER2+ metastatic colorectal cancer and established the role of HER2 as a clinically actionable target (1–5). On the basis of these results, trastuzumab plus lapatinib or pertuzumab regimens have been included in the National Comprehensive Cancer Network guidelines and trastuzumab plus tucatinib has been recently granted accelerated approval by the FDA. However, a substantial proportion of patients enrolled in clinical trials did not benefit from these targeted strategies, as the rate of early disease progression is 25% to 41% and the median progression-free survival (PFS) ranges from 2.9 to 8.2 months according to the specific regimen and the adopted molecular selection criteria (2, 3, 5, 6). The biological bases of primary resistance to dual HER2 targeting in patients with HER2+ metastatic colorectal cancer are poorly characterized. The cooccurrence of other oncogenic drivers is mostly represented by RAS comutations, reported in around 17% of patients (7). Because (K)RAS mutations have been clearly associated with extremely poor outcomes after dual HER2 blockade, the most recent trials restricted the enrolment to patients with RAS wild-type (WT) status (2, 6).

Tissue and circulating tumor DNA (ctDNA) exploratory analyses of phase II trials suggested that additional uncommon alterations in receptor tyrosine kinase (RTK)/MAPK pathway may bypass therapeutic HER2 blockade. Therefore, a paradigm of negative selection beyond RAS status is needed to potentially improve the precision of HER2-targeting strategies. In addition, tissue or ctDNA HER2 copy-number variation (CNV) assessed by RT-PCR or next-generation sequencing (NGS) has been associated with the outcomes of dual HER2 blockade (1, 3, 6). In fact, the level of HER2 amplification in tumor cells may be a surrogate of HER2 addiction and predict the sensitivity to trastuzumab-based regimens. Of note, comprehensive genomic profiling (CGP) may concomitantly identify the potential drivers of primary resistance and HER2 CNV.

Drawing from these considerations, we hypothesized that genomic-based hyper-selection may refine the prognostic stratification of patients receiving dual HER2 blockade. To this aim, we conducted a multinational effort aimed at investigating the prognostic performance of a panel of rare genomic drivers of primary resistance (i.e., the PRESSING-HER2 panel) and HER2 CNV, both assessed by NGS, in patients with HER2+ and RAS WT metastatic colorectal cancer receiving dual HER2 blockade.

Patient population

Patients with HER2+, RAS WT, and microsatellite stable metastatic colorectal cancer treated with trastuzumab-based dual HER2 blockade were retrieved from three different screening sources (Supplementary Table S1): a prospective observational study in Italy and Spain, the MSK-IMPACT dataset, and the TRIUMPH trial (3, 8–10). Additional inclusion criteria were: availability of CGP data, Eastern Cooperative Oncology Group (ECOG) Performance Status (PS) ≤ 2, at least one measurable lesion according to RECIST v1.1, at least one prior treatment line for metastatic disease and written informed consent to study participation. HER2 positivity was defined by: (i) HER2 IHC 3+ in ≥ 10% of cells or HER2 IHC 2+ and HER2/CEP17 ratio ≥ 2 by ISH according to previously reported criteria (1); or (ii) presence of HER2 amplification detected by NGS and defined by HER2 CNV ≥6. Response assessment was performed according to RECIST v1.1 and CT scans were performed every 8 ± 1 weeks. Primary resistance to dual HER2 blockade was defined by PFS <4 months and best response stable disease (SD) or progressive disease (PD), whereas sensitivity was defined by PFS ≥4 months regardless of RECIST response, that is, SD, partial response (PR - including unconfirmed PR) or complete response (CR). The study was approved by the Institutional Review Board of the Fondazione IRCCS Istituto Nazionale dei Tumori di Milano (INT 117/15) and was conducted in accordance with the ethical principles for medical research involving human subjects adopted in the Declaration of Helsinki.

CGP

CGP was performed in archival formalin-fixed paraffin-embedded tumor tissue obtained prior to the start of anti-HER2 therapy. The PRESSING-HER2 panel grouped genomic alterations with sound clinical and biological rationale as driver of primary resistance to anti-HER2 blockade and included: (i) HER2 on-target alterations, that is, HER2 pathogenic mutations or rearrangements shown to drive resistance; (ii) off-target alterations, that is, mutations/amplifications in RTK/MAPK genes: EGFR, MET, or KRAS coamplifications; BRAF class 1 and 2 mutations or PIK3CA exon 20 mutations. Patients receiving trastuzumab plus pertuzumab were considered as PRESSING-HER2+ if harboring HER2 mutations in the tyrosine kinase domain and established as resistant to trastuzumab plus pertuzumab in the TAPUR trial, such as: L755S, R678Q, L755-T759, D769H, D769Y, V777L, P780ins, V842I, R896C, or HER2 pathogenic fusions (4, 11). Regarding patients receiving trastuzumab plus a tyrosine kinase inhibitor (TKI), HER2 mutations were included in the PRESSING-HER2 panel according to the robust literature data on their role as driver of resistance to the specific TKI used (12–15). Regarding HER2 CNV assessed by NGS, tumors were reclassified as nonamplified if HER2 CNV was <6 despite HER2 positivity initially detected by IHC +/− ISH.

Statistical analysis

The study was designed as a multicenter, case–control study based on a translational hypothesis of primary resistance to dual HER2 inhibition. The independent collection of cases (primary resistant group) and controls (sensitive group) with one control per case was planned. In the TRIUMPH trial Nakamura reported the presence of three driver resistance mechanisms out of 7 patients with PD as best response to trastuzumab and pertuzumab (42%) as assessed by archival tissue NGS (3). Therefore, we hypothesized that a more conservative prevalence threshold of 30% would better apply to the present definition of primary resistance. Therefore, hypothesizing a prevalence of PRESSING-HER2 alterations equal to 30% and 5% among cases and controls, respectively, 35 patients per group were needed to reject the null hypothesis of equally prevalent alterations, with α and β errors of 0.05 and 0.20. An uncorrected χ2 statistic was used to compare the prevalence of alterations in the PRESSING-HER2 panel and in other alterations between resistant and sensitive patients. PFS was defined as the time from the start of dual anti-HER2 treatment to disease progression or death from any cause. Overall survival (OS) was defined as the time from the start of dual anti-HER2 treatment to death from any cause or last follow-up for alive patients. The Kaplan–Meier estimator and Cox proportional hazards regression were used for survival analysis using the survival, survminer, and survMisc packages of the R software (version 3.5.0) and R Studio (version 2022.07.2). In Cox proportional hazards regression models, all the covariates associated with PFS and OS in the univariable analyses with a P < 0.05 were included in the multivariable model. P < 0.05 was considered statistically significant. HER2 CNV was modeled by means of 3-knots natural cubic splines (using the splines package) to assess flexible fit and to check for nonlinearity.

Data availability

The data generated during and/or analyzed during this study are available within the article and its Supplementary Files. Genomic data, including CNVs, were pulled from other sources and can be requested to the sources detailed in Supplementary Table S1; scientific agreements would be required, and the corresponding author can provide the report ID numbers upon reasonable request.

Study population

The study flowchart is depicted in Supplementary Fig. S1. The final study population included 33 cases and 35 controls. HER2 positivity was identified by IHC +/− ISH in 54 (79%) cases and by NGS alone in 14 (21%). Among the screened patients, no RECIST PR were observed in those with PFS <4 months. HER2 amplification was confirmed by NGS in 43 of 45 patients with IHC 3+ and only in 4 of 9 with IHC 2+, with overall prevalence of HER2 nonamplified samples of 10%. Table 1 summarizes the patient and disease characteristics, overall and according to the status of primary resistance versus sensitivity. No statistically significant differences in terms of the main baseline variables were observed between the two resistant and sensitive cohorts, except for ECOG PS. In fact, the prevalence of ECOG PS 1–2 was 64% versus 17% (P < 0.001) in resistant versus sensitive groups. Eighteen patients (27%) had received prior anti-EGFR–based therapy, without significant differences in resistant versus sensitive groups (24% vs. 29%; P = 0.786). Trastuzumab was given in combination with pertuzumab in 44 (65%) patients, or with the TKI lapatinib, tucatinib, or neratinib in 24 (35%). In the overall population, at a median follow-up of 18.7 months (IQR, 13.4–27.3), the median PFS was 4.1 months (95% CI, 2.8–5.5), and the median OS was 12.2 months (95% CI, 9.9–16.4), as shown in Supplementary Fig. S2.

Table 1.

Patients and disease baseline characteristics, overall and according to primary resistance vs. sensitivity to dual HER2 blockade.

CharacteristicsOverall study population (N = 68) N (%)Resistant patients (n = 33) n (%)Sensitive patients (n = 35) n (%)P
Age (years) — — — >0.999 
 <70 55 (81) 27 (82) 28 (80) — 
 ≥70 13 (19) 6 (18) 7 (20) — 
Sex — — — 0.811 
 Female 38 (56) 19 (58) 19 (54) — 
 Male 30 (44) 14 (42) 16 (46) — 
ECOG PS — — — <0.001 
 0 41 (60) 12 (36) 29 (83) — 
 1–2 27 (40) 21 (64) 6 (17) — 
Primary tumor location — — — 0.349 
 Right colon 4 (6) 3 (9) 1 (3) — 
 Left colon/Rectum 64 (94) 30 (91) 34 (97) — 
Primary tumor resection — — — >0.999 
 Yes 14 (21) 26 (79) 28 (80) — 
 No 54 (79) 7 (21) 7 (20) — 
Prior adjuvant chemotherapy — — — 0.806 
 Yes 26 (38) 21 (64) 21 (60) — 
 No 42 (62) 12 (36) 14 (40) — 
Metastatic sites (N— — — 0.580 
 1 17 (25) 7 (21) 10 (29) — 
 >1 51 (75) 26 (79) 25 (71) — 
Prior exposure to anti-EGFR — — — 0.786 
 Yes 18 (27) 8 (24) 10 (29) — 
 No 50 (73) 25 (76) 25 (71) — 
Prior treatment lines (N— — — 0.341 
 1–2 31 (46) 13 (39) 18 (51) — 
 ≥3 37 (54) 20 (61) 17 (49) — 
Anti-HER2 therapy — — — 0.079 
 mAbs 44 (65) 25 (76) 19 (54) — 
 TKI plus trastuzumab 24 (35) 8 (24) 16 (46) — 
CharacteristicsOverall study population (N = 68) N (%)Resistant patients (n = 33) n (%)Sensitive patients (n = 35) n (%)P
Age (years) — — — >0.999 
 <70 55 (81) 27 (82) 28 (80) — 
 ≥70 13 (19) 6 (18) 7 (20) — 
Sex — — — 0.811 
 Female 38 (56) 19 (58) 19 (54) — 
 Male 30 (44) 14 (42) 16 (46) — 
ECOG PS — — — <0.001 
 0 41 (60) 12 (36) 29 (83) — 
 1–2 27 (40) 21 (64) 6 (17) — 
Primary tumor location — — — 0.349 
 Right colon 4 (6) 3 (9) 1 (3) — 
 Left colon/Rectum 64 (94) 30 (91) 34 (97) — 
Primary tumor resection — — — >0.999 
 Yes 14 (21) 26 (79) 28 (80) — 
 No 54 (79) 7 (21) 7 (20) — 
Prior adjuvant chemotherapy — — — 0.806 
 Yes 26 (38) 21 (64) 21 (60) — 
 No 42 (62) 12 (36) 14 (40) — 
Metastatic sites (N— — — 0.580 
 1 17 (25) 7 (21) 10 (29) — 
 >1 51 (75) 26 (79) 25 (71) — 
Prior exposure to anti-EGFR — — — 0.786 
 Yes 18 (27) 8 (24) 10 (29) — 
 No 50 (73) 25 (76) 25 (71) — 
Prior treatment lines (N— — — 0.341 
 1–2 31 (46) 13 (39) 18 (51) — 
 ≥3 37 (54) 20 (61) 17 (49) — 
Anti-HER2 therapy — — — 0.079 
 mAbs 44 (65) 25 (76) 19 (54) — 
 TKI plus trastuzumab 24 (35) 8 (24) 16 (46) — 

Abbreviations: ECOG PS, Eastern Cooperative Oncology Group performance status; mAb, monoclonal antibody; TKI, tyrosine-kinase inhibitor.

CGP results

The genomic profiles per single patient in the two cohorts of resistant versus sensitive tumors are depicted in the heat map in Fig. 1. PRESSING-HER2 panel alterations were detected in 10 (15%) patients and were mutually exclusive of each other. The following alterations were detected: HER2 pathogenic mutations occurring in the protein tyrosine and serine/threonine kinase domain (i.e., V777L, V842I, D769Y, and V777_Gly778insGSP) in five tumor samples, HER2 rearrangements in two samples, EGFR coamplification in one sample, and BRAF class 1 or 2 mutations (i.e., V600E and G469A) in two samples. No MET or KRAS coamplifications were detected. A significantly higher frequency of PRESSING-HER2 alterations was found in resistant (9 of 33, 27%) versus sensitive patients (1 of 35, 3% - P = 0.005). The accuracy of the PRESSING-HER2 panel for predicting the status of primary resistance was 63%. The individual features and treatment outcomes of patients with PRESSING-HER2+ tumors are detailed in Supplementary Table S2. Regarding patients with HER2-mutated tumors, 4 received trastuzumab plus pertuzumab, whereas 3 received trastuzumab plus a TKI. Among these, a patient treated with trastuzumab plus lapatinib harbored the lapatinib-resistant V482I mutation and it was therefore considered PRESSING-HER2+ (16). The remaining patients were considered a priori as PRESSING-HER2, despite being clinically resistant in our dataset: the first received trastuzumab plus lapatinib and harbored the lapatinib-sensitive H878Y mutation (17); the second received trastuzumab plus neratinib and harbored the neratinib-sensitive D769Y mutation (15). Two patients with HER2-rearranged tumors received trastuzumab plus pertuzumab; the first harbored the GRB7-HER2 fusion (HER2 CNV 198) and was in the resistant cohort, while the second patient with WIPF2-HER2 (HER2 CNV 163) fusion was classified as sensitive, but had only a 4.4-month lasting SD. The median HER2 CNV was 34.5 (IQR, 18.7–78.5) in the overall population, without significant differences in such value according to the three NGS assays used. Notably, median HER2 CNV was 23.0 (IQR, 9.0–41.4) versus 68.5 (IQR, 33.4–105.0) in the resistant versus sensitive cohort (P <0.001; Supplementary Fig. S3). On the other hand, median CNV was 31.7 (IQR, 13.2–63.2) in PRESSING-HER2+ versus 35 (IQR, 19.0–78.5) in PRESSING-HER2 tumors (P = 0.489); HER2 CNV <6 was found in 20% (2 of 10) of PRESSING-HER2+ versus 9% of PRESSING-HER2 tumors (5 of 58; P = 0.272). When combining the presence of PRESSING-HER2 with lack of HER2 amplification by NGS (HER2 CNV <6), the predictive accuracy of primary resistance was 70%. In an exploratory analysis restricting the sensitive group to patients with CR/PR as RECIST best response, a significantly higher frequency of PRESSING-HER2 alterations was found in resistant (9 of 33, 27%) versus sensitive patients (0 of 19, 0%; P = 0.018).

Figure 1.

Heat map showing the genomic profiles according to the primary resistance status. Patients in the two groups were ordered according to the length of individual PFS. CNV, copy number variation.

Figure 1.

Heat map showing the genomic profiles according to the primary resistance status. Patients in the two groups were ordered according to the length of individual PFS. CNV, copy number variation.

Close modal

Prognostic role of PRESSING-HER2 panel and HER2 CNV

Patients with PRESSING-HER2+ tumors had significantly worse PFS and OS compared with those with PRESSING-HER2 status (median PFS: 2.2 vs. 5.3 months; HR, 3.76; 95% CI, 1.78–7.95; P < 0.001; median OS: 5.4 vs. 14.9 months; HR, 3.61; 95% CI, 1.66–7.82; P = 0.001; Fig. 2A and B). Regarding HER CNV as a continuous variable, a linear effect on the log hazard function was observed for PFS (P for nonlinearity = 0.142) and OS (P for nonlinearity = 0.884) as shown in Supplementary Fig. S4. Compared with patients with HER2-amplified status confirmed by NGS (HER2 CNV ≥6), patients with HER2 nonamplified tumors (HER2 CNV <6) had significantly worse PFS (median PFS: 1.6 vs. 5.2 months; HR, 4.63; 95% CI, 1.97–10.92; P < 0.001) and nonsignificantly inferior OS (median OS, 7.4 vs. 12.4 months; HR, 1.87; 95% CI, 0.79–4.46; P = 0.157), as shown in Fig. 2C and D. In the multivariable models (Table 2), the presence of PRESSING-HER2 alterations was significantly associated with both PFS and OS (adjusted HR for PFS = 3.06, 95% CI, 1.40–6.69, P = 0.005; adjusted HR for OS = 2.93, 95% CI, 1.32–6.48, P = 0.007), whereas HER2 nonamplified status by NGS was significantly associated only with PFS (adjusted HR, 3.89; 95% CI, 1.60–9.49; P = 0.002), but not with OS (HR, 1.87; 95% CI, 0.79–4.46; P = 0.157). Notably, ECOG PS was independently associated with both PFS and OS.

Figure 2.

Kaplan–Meier curves for PFS (A and C) and OS (B and D) according to the presence of PRESSING-HER2 status (positive vs. negative) and HER2 amplification status by NGS (HER2 CNV <6 vs. ≥6). mOS, median overall survival; mPFS, median progression-free survival; P-HER2, PRESSING-HER2 panel.

Figure 2.

Kaplan–Meier curves for PFS (A and C) and OS (B and D) according to the presence of PRESSING-HER2 status (positive vs. negative) and HER2 amplification status by NGS (HER2 CNV <6 vs. ≥6). mOS, median overall survival; mPFS, median progression-free survival; P-HER2, PRESSING-HER2 panel.

Close modal
Table 2.

Cox proportional hazards regression models for PFS and OS in the entire study population.

PFSOS
Univariable modelsMultivariable modelUnivariable modelsMultivariable model
CharacteristicsHR (95% CI)PHR (95% CI)PHR (95% CI)PHR (95% CI)P
Age (years) — 0.412 — — — 0.813 — — 
 <70 Ref — — — Ref — — — 
 ≥70 1.31 (0.69–2.48) — — — 1.09 (0.52–2.28) — — — 
Sex — 0.798 — — — 0.508 — — 
 Female Ref — — — Ref — — — 
 Male 0.93 (0.56–1.56) — — — 0.81 (0.44–1.50) — — — 
ECOG PS — 0.001 — 0.026 — <0.001 — <0.001 
 0 Ref — Ref — Ref — Ref — 
 ≥1 2.39 (1.42–4.03) — 1.88 (1.08–3.29) — 3.32 (1.82–6.06) — 3.04 (1.65–5.59) — 
Primary tumor location — 0.247 — — — 0.630 — — 
 Right Ref — — — Ref — — — 
 Left colon/Rectum 0.54 (0.19–1.52) — — — 0.75 (0.23–2.43) — — — 
Primary tumor resection — 0.498 — — — 0.453 — — 
 No Ref — — — Ref — — — 
 Yes 0.80 (0.41–1.54) — — — 1.31 (0.64–2.69) — — — 
Adjuvant CT — 0.449 — — — 0.560 — — 
 No Ref — — — Ref — — — 
 Yes 1.22 (0.73–2.06) — — — 0.83 (0.44–1.56) — — — 
Metastatic sites (N— 0.487 — — — 0.218 — — 
 1 Ref — — — Ref — — — 
 >1 1.23 (0.68–2.22) — — — 1.62 (0.75–3.51) — — — 
Prior treatment lines — 0.288 — — — 0.065 — — 
 1–2 Ref — — — Ref — — — 
 ≥3 1.32 (0.79–2.20) — — — 1.79 (0.96–3.31) — — — 
PRESSING-HER2 — <0.001 — 0.005 — <0.001 — 0.007 
 Negative Ref — Ref — Ref — Ref — 
 Positive 3.49 (1.74–6.99) — 3.06 (1.40–6.69) — 3.79 (1.78–8.07) — 2.93 (1.32–6.48) — 
HER CNV — <0.001 — 0.002 — 0.157 — — 
 ≥6 Ref — Ref — Ref — — — 
 <6 4.63 (1.97–10.92) — 3.89 (1.60–9.49) — 1.87 (0.79–4.46) — — — 
PFSOS
Univariable modelsMultivariable modelUnivariable modelsMultivariable model
CharacteristicsHR (95% CI)PHR (95% CI)PHR (95% CI)PHR (95% CI)P
Age (years) — 0.412 — — — 0.813 — — 
 <70 Ref — — — Ref — — — 
 ≥70 1.31 (0.69–2.48) — — — 1.09 (0.52–2.28) — — — 
Sex — 0.798 — — — 0.508 — — 
 Female Ref — — — Ref — — — 
 Male 0.93 (0.56–1.56) — — — 0.81 (0.44–1.50) — — — 
ECOG PS — 0.001 — 0.026 — <0.001 — <0.001 
 0 Ref — Ref — Ref — Ref — 
 ≥1 2.39 (1.42–4.03) — 1.88 (1.08–3.29) — 3.32 (1.82–6.06) — 3.04 (1.65–5.59) — 
Primary tumor location — 0.247 — — — 0.630 — — 
 Right Ref — — — Ref — — — 
 Left colon/Rectum 0.54 (0.19–1.52) — — — 0.75 (0.23–2.43) — — — 
Primary tumor resection — 0.498 — — — 0.453 — — 
 No Ref — — — Ref — — — 
 Yes 0.80 (0.41–1.54) — — — 1.31 (0.64–2.69) — — — 
Adjuvant CT — 0.449 — — — 0.560 — — 
 No Ref — — — Ref — — — 
 Yes 1.22 (0.73–2.06) — — — 0.83 (0.44–1.56) — — — 
Metastatic sites (N— 0.487 — — — 0.218 — — 
 1 Ref — — — Ref — — — 
 >1 1.23 (0.68–2.22) — — — 1.62 (0.75–3.51) — — — 
Prior treatment lines — 0.288 — — — 0.065 — — 
 1–2 Ref — — — Ref — — — 
 ≥3 1.32 (0.79–2.20) — — — 1.79 (0.96–3.31) — — — 
PRESSING-HER2 — <0.001 — 0.005 — <0.001 — 0.007 
 Negative Ref — Ref — Ref — Ref — 
 Positive 3.49 (1.74–6.99) — 3.06 (1.40–6.69) — 3.79 (1.78–8.07) — 2.93 (1.32–6.48) — 
HER CNV — <0.001 — 0.002 — 0.157 — — 
 ≥6 Ref — Ref — Ref — — — 
 <6 4.63 (1.97–10.92) — 3.89 (1.60–9.49) — 1.87 (0.79–4.46) — — — 

Abbreviations: CT, chemotherapy; Ref, reference.

In the combined assessment of PRESSING-HER2 panel and HER2 amplification assessed by NGS, patients with PRESSING-HER2+ and/or HER2 nonamplified tumors had significantly worse PFS and OS compared with patients without PRESSING-HER2 alterations and HER2 CNV ≥6 (median PFS: 2.1 vs. 5.5 months; HR = 4.70, 95% CI, 2.35–9.41; P < 0.001; median OS = 6.7 vs. 14.9 months, HR = 2.69, 95% CI, 1.38–5.23; P = 0.004; Fig. 3A and B).

Figure 3.

Kaplan–Meier curves for PFS (A) and OS (B) according to the combined assessment of PRESSING-HER2 and HER2 amplification status by NGS (PRESSING-HER2 and HER2 amplified vs. PRESSING-HER2+ and/or HER2 nonamplified). mOS, median overall survival; mPFS, median progression-free survival; P-HER2, PRESSING-HER2 panel.

Figure 3.

Kaplan–Meier curves for PFS (A) and OS (B) according to the combined assessment of PRESSING-HER2 and HER2 amplification status by NGS (PRESSING-HER2 and HER2 amplified vs. PRESSING-HER2+ and/or HER2 nonamplified). mOS, median overall survival; mPFS, median progression-free survival; P-HER2, PRESSING-HER2 panel.

Close modal

In nonrandomized trials, dual HER2 blockade showed promising activity in patients with previously treated HER2+ metastatic colorectal cancer. However, the recently reported SWOG S1613 randomized phase II trial failed to show the superiority of a trastuzumab plus pertuzumab chemo-free regimen compared with irinotecan plus cetuximab as second- or third-line treatment in patients with HER2+, RAS and BRAF WT disease (18). Although these results may have been influenced by the small sample size and the lack of prior irinotecan exposure in about half of the patients, this study confirms that a nonnegligible proportion of patients do not derive benefit from anti-HER2–targeted therapy. Therefore, it is critical to identify the determinants of treatment resistance in this molecularly selected subgroup of patients characterized by high degree of genomic heterogeneity. Because of the low prevalence of each candidate resistance alteration and the current lack of large randomized clinical trials with an anti-HER2–free arm, a formal evaluation of the negative predictive role of these mechanisms individually is not currently feasible. In the attempt to partially overcome these limitations, we investigated the role of a genomic panel that groups together several uncommon biomarkers of primary resistance in a case–control study on the basis of a formal a priori statistical hypothesis. We found that genomic alterations included in the PRESSING-HER2 panel were significantly enriched in patients exhibiting primary resistance to dual HER2 inhibition and were independently associated with inferior PFS and OS in multivariable analyses.

Expectedly, all patients bearing HER2 off-target alterations (RTK/MAPK pathway) were resistant irrespective of the specific anti-HER2 regimen with trastuzumab plus either pertuzumab or a TKI. Among patients with HER2 on-target alterations, those with HER2 activating mutations treated with trastuzumab plus pertuzumab were expectedly resistant, whereas patients treated with a TKI were classified as PRESSING-HER2 positive or negative based on the available preclinical and/or clinical literature data regarding the specific agent (14, 15). Indeed, there is evidence indicating that several HER2 activating mutations in colorectal cancer (S310F, L755S, V777L, V842I, and L866M) are efficiently targeted by tucatinib or neratinib plus trastuzumab, whereas the binding of lapatinib might be impaired in presence of V842I and L755S; refs. 15, 16, 19). Although all patients with HER2 mutations in our dataset were in the resistant cohort irrespective of initial classification as PRESSING-HER2 positive or negative, conclusive data regarding the predictive impact of individual HER2 mutations in the context of HER2-amplified tumors are lacking. The same is true for HER2 rearrangements: in fact, gene fusions may cause constitutive kinase activity and resistance to extracellular domain–targeting agents (11, 20). Of note, similarly to the reported occurrence of EGFR fusions in EGFR-amplified metastatic colorectal cancer, HER2 rearrangements occurred in HER2 “hyperamplified” tumors with very high CNV (9).

A nonnegligible proportion (10%) of patients in our dataset did not show HER2 amplification by NGS despite HER2 positivity having been previously detected by standard IHC ± ISH. The lack of HER2 amplification by NGS may be a consequence of spatial heterogeneity of HER2 amplification and may thus mirror a low level of HER2 addiction. Accordingly, we showed that patients with HER2 nonamplified tumors by NGS had poorer outcomes, although the low number of patients in this subgroup may have prevented significant OS results. Interestingly, most patients with HER2 nonamplified tumors by NGS did not have concomitant PRESSING-HER2 alterations, and therefore the combined assessment of the two biomarkers increased the predictive accuracy and the prognostic stratification of the survival outcomes.

Herein, poorer ECOG PS was significantly associated with survival, regardless of PRESSING-HER2 alterations or HER2 nonamplified status by NGS. These data strengthen the need of implementing HER2 blockade in earlier treatment lines, ideally before the potential occurrence of ECOG PS decline in the chemorefractory setting.

Collectively, our data suggest that CGP may improve the negative selection for anti-HER2 therapy by the concomitant assessment of resistance alterations and HER2 CNV. However, the definitive demonstration of the clinical usefulness of GCP in this setting should derive from the validation of the negative predictive role of the above discussed biomarkers in the context of randomized clinical trials with an anti-HER2–free arm, such as the ongoing MOUNTAINEER-3 (21). Moreover, our study was focused on primary resistance to anti-HER2 therapy, but HER2 CNV as measured by CGP may also allow to identify HER2 hyperamplified tumors with exceptional benefit from dual HER2 blockade, as suggested by previous translational analyses of clinical trials (3, 6, 18).

Consistent with the experience of EGFR blockade, patients with PIK3CA mutations or right-sided primary tumor had inferior outcomes on trastuzumab plus pertuzumab in the MyPathway trial, that included HER2+ patients regardless of RAS mutational status (2). However, because both PIK3CA mutations and right sidedness are associated with RAS mutations in metastatic colorectal cancer, their prognostic role in patients with HER2+ and RAS WT tumors is largely undetermined (22). After the exclusion of RAS mutated samples in our dataset, we did not report any PIK3CA exon 20 mutation and only 4 (6%) right-sided tumors. Thus, we could not draw any conclusion on the prognostic impact of primary tumor sidedness in the setting of HER2 inhibition. Similarly, BRAFV600E mutations have a clear-cut role in mediating resistance to RTK pathway blockade strategies. However, only individual patients with BRAFV600E metastatic colorectal cancer and resistant disease have been reported in two clinical trials with dual HER2 blockade, and our results seem to confirm the negative prognostic role of cooccurring BRAF alterations (2, 3).

Notably, we previously showed that the antibody–drug conjugate trastuzumab deruxtecan (T-DXd) may bypass specific drivers of primary resistance (such as KRAS amplification) in HER2-amplified gastric cancer models treated with several anti-HER2 combinations (23). More importantly, the exploratory analyses of the DESTINY-CRC-01 trial showed a potentially retained activity of this agent in patients with RAS mutations or HER2 CNV-low status in ctDNA (24). Therefore, T-DXd may be regarded as a “smart chemotherapy” option and may be active irrespective of the presence of genomic resistance alterations. Our results suggest that treatment decisions and sequencing of the available anti-HER2 options may consider the results of CGP, but its potential role in driving treatment choices should be interpreted with caution. More preclinical/translational data are needed to comprehensively investigate sensitivity to different HER2 targeted therapies, and additional trials or real-world evidence are necessary.

Our study has several limitations. First, we acknowledge that the definition of sensitive patients by PFS ≥4 months and no PD as best RECIST response (especially those not achieving PR/CR) may have led to the inclusion of patients with indolent disease rather than truly anti-HER2s sensitive in this group. However, in this hard-to-treat heavily preteated patient population, even a short-lasting disease stabilization may be considered as a sign of treatment efficacy. Moreover, we showed that the discriminative capability of the PRESSING-HER2 panel was preserved after excluding patients with SD as their best response in the sensitive cohort.

Then, while we cannot conclusively demonstrate the predictive role of the biomarkers tested, the use of a case–control study design allowed us to closely mirror the setting of a predictive validation. Second, the reproducibility of HER2 CNV assessment may have been negatively influenced by the heterogeneity of the NGS assays. It should be also kept in mind that NGS may underestimate the HER2 CNV because of the stromal dilution as compared with standard morphologic assays such as ISH (25). Finally, a substantial proportion of patients with primary resistance did not display any PRESSING-HER2 alteration nor low HER2 CNV by tissue NGS. As a matter of fact, heterogeneity of genomic profiles as well as nongenomic resistance may account for primary resistance to the dual HER2 blockade (24, 26). Therefore, liquid biopsy may further improve the stratification of patients’ outcomes, as previously shown in the TRIUMPH and HERACLES studies (1, 3).

In conclusion, a panel of genomic on/off target resistance alterations and the lack of HER2 amplification as assessed by NGS may be useful to predict primary resistance to dual HER2 blockade in patients with HER2+ and RAS WT metastatic colorectal cancer. CGP may allow physicians to refine patients’ selection through negative hyperselection beyond RAS mutational status applied to the context of HER2 overexpression/amplification as a positive predictive biomarker.

Y. Nakamura reports grants and personal fees from Chugai; grants from Guardant Health, Roche Diagnostics, Taiho, Daiichi Sankyo, Genomedia, and Seagen; and personal fees from Merck Biopharma and Guardant Health AMEA outside the submitted work. R. Yaeger reports grants and personal fees from Mirati Therapeutics; grants from Pfizer, Boehringer Ingelheim, and Daiichi Sankyo; and personal fees from Zai Lab outside the submitted work. S. Lonardi reports personal honoraria as invited speaker from Roche, Eli Lilly, Bristol Myers Squibb, Servier, Merck Serono, Pierre Fabre, GlaxoSmithKline, and Amgen and participation in advisory board for Amgen, Astellas, Bayer, Merck Serono, Eli Lilly, AstraZeneca, Incyte, Daiichi-Sankyo, Bristol Myers Squibb, Servier, Merck Sharp & Dohme, GlaxoSmithKline, and Takeda. C. Cremolini reports grants and personal fees from Servier, Amgen, Merck, and Bayer and personal fees from Roche, Mirati, Nordic Pharma, Pierre Fabre, and MSD outside the submitted work. E. Elez reports personal fees from Amgen, Merck Serono, Seagen, Bayer, Servier, Organon, Pierre Fabre, Takeda, MSD, and Sanofi Aventis outside the submitted work. F. Bergamo reports personal fees from Servier, AAA Novartis, Eli Lilly, MSD, EISAI, Bayer, and BMS outside the submitted work. H. Bando reports other support from Eli Lilly Japan, Taiho Pharmaceutical, and Ono Pharmaceutical outside the submitted work. A. Raimondi reports other support from Servier, Elma Academy, and Amgen outside the submitted work. A.B. Schrock reports Foundation Medicine and other support from Roche during the conduct of the study. T. Yoshino reports grants from Taiho, Amgen, Daiichi-Sankyo, Eisai, FALCO biosystems, Genomedia, Molecular Health, Nippon Boehringer Ingelheim, Pfizer, Roche Diagnostics, Sysmex, and Sanofi; grants and personal fees from Ono, Chugai, and MSD; personal fees from Bayer, Merck Biopharma, and Takeda; and other support from Sumitomo Corp. outside the submitted work. F. Pietrantonio reports grants and personal fees from Amgen and BMS; grants from Incyte, AstraZeneca, and Agenus; and personal fees from Bayer, Pierre-Fabre, Servier, Takeda, MSD, Astellas, and Merck Serono outside the submitted work. No disclosures were reported by the other authors.

G. Randon: Conceptualization, data curation, software, formal analysis, investigation, visualization, methodology, writing–original draft. Y. Nakamura: Resources, investigation. R. Yaeger: Resources, data curation, investigation, writing–review and editing. S. Lonardi: Resources, data curation, investigation, writing–review and editing. C. Cremolini: Resources, data curation, investigation, writing–review and editing. E. Elez: Resources, data curation, investigation, writing–review and editing. F. Nichetti: Data curation, software, formal analysis, investigation, methodology, writing–original draft. F. Ghelardi: Data curation, investigation, writing–original draft. V. Nasca: Data curation, investigation, writing–original draft. F. Bergamo: Resources, data curation, investigation, writing–review and editing. V. Conca: Resources, data curation, investigation, writing–review and editing. J. Ros: Resources, data curation, investigation, writing–review and editing. H. Bando: Resources, data curation, investigation, writing–review and editing. G. Maddalena: Resources, data curation, investigation, writing–review and editing. S. Oldani: Data curation, investigation. M. Prisciandaro: Resources, data curation, investigation, writing–original draft. A. Raimondi: Resources, data curation, investigation, writing–original draft. A.B. Schrock: Resources, data curation, investigation, writing–review and editing. L. Agnelli: Resources, data curation, software, formal analysis, investigation, methodology. H. Walch: Resources, data curation, software, formal analysis, investigation, methodology. T. Yoshino: Resources, data curation, investigation, writing–review and editing. F. Pietrantonio: Conceptualization, resources, data curation, supervision, funding acquisition, investigation, visualization, methodology, writing–original draft, project administration.

This study was supported by AIRC IG 23624 (to F. Pietrantonio), and by the NIH Cancer Center Core Grant P30 CA008748 to Memorial Sloan Kettering Cancer Center.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.

Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).

1.
Sartore-Bianchi
A
,
Trusolino
L
,
Martino
C
,
Bencardino
K
,
Lonardi
S
,
Bergamo
F
, et al
.
Dual-targeted therapy with trastuzumab and lapatinib in treatment-refractory, KRAS codon 12/13 wild-type, HER2-positive metastatic colorectal cancer (HERACLES): a proof-of-concept, multicentre, open-label, phase 2 trial
.
Lancet Oncol
2016
;
17
:
738
46
.
2.
Meric-Bernstam
F
,
Hurwitz
H
,
Raghav
KPS
,
McWilliams
RR
,
Fakih
M
,
VanderWalde
A
, et al
.
Pertuzumab plus trastuzumab for HER2-amplified metastatic colorectal cancer (MyPathway): an updated report from a multicentre, open-label, phase 2a, multiple basket study
.
Lancet Oncol
2019
;
20
:
518
30
.
3.
Nakamura
Y
,
Okamoto
W
,
Kato
T
,
Esaki
T
,
Kato
K
,
Komatsu
Y
, et al
.
Circulating tumor DNA-guided treatment with pertuzumab plus trastuzumab for HER2-amplified metastatic colorectal cancer: a phase 2 trial
.
Nat Med
2021
;
27
:
1899
903
.
4.
Gupta
R
,
Meric-Bernstam
F
,
Rothe
M
,
Garrett-Mayer
E
,
Mangat
PK
,
D'Andre
S
, et al
.
Pertuzumab plus trastuzumab in patients with colorectal cancer with ERBB2 amplification or ERBB2/3 mutations: results from the TAPUR study
.
JCO Precis Oncol
2022
;
6
:
e2200306
.
5.
Strickler
JH
,
Cercek
A
,
Siena
S
,
André
T
,
Ng
K
,
Van Cutsem
E
, et al
.
Tucatinib plus trastuzumab for chemotherapy-refractory, HER2-positive, RAS-wild-type unresectable or metastatic colorectal cancer (MOUNTAINEER): a multicentre, open-label, phase 2 study
.
Lancet Oncol
2023
;
24
:
496
508
.
6.
Siravegna
G
,
Sartore-Bianchi
A
,
Nagy
RJ
,
Raghav
K
,
Odegaard
JI
,
Lanman
RB
, et al
.
Plasma HER2 (ERBB2) copy number predicts response to HER2-targeted therapy in metastatic colorectal cancer
.
Clin Cancer Res
2019
;
25
:
3046
53
.
7.
Ross
JS
,
Fakih
M
,
Ali
SM
,
Elvin
JA
,
Schrock
AB
,
Suh
J
, et al
.
Targeting HER2 in colorectal cancer: the landscape of amplification and short variant mutations in ERBB2 and ERBB3
.
Cancer
2018
;
124
:
1358
73
.
8.
Chatila
WK
,
Kim
JK
,
Walch
H
,
Marco
MR
,
Chen
C-T
,
Wu
F
, et al
.
Genomic and transcriptomic determinants of response to neoadjuvant therapy in rectal cancer
.
Nat Med
2022
;
28
:
1646
55
.
9.
Randon
G
,
Yaeger
R
,
Hechtman
JF
,
Manca
P
,
Fucà
G
,
Walch
H
, et al
.
EGFR amplification in metastatic colorectal cancer
.
J Natl Cancer Inst
2021
;
113
:
1561
9
.
10.
Elez
E
,
Ros
J
,
Fernández
J
,
Villacampa
G
,
Moreno-Cárdenas
AB
,
Arenillas
C
, et al
.
RNF43 mutations predict response to anti-BRAF/EGFR combinatory therapies in BRAFV600E metastatic colorectal cancer
.
Nat Med
2022
;
28
:
2162
70
.
11.
Hechtman
JF
,
Zehir
A
,
Yaeger
R
,
Wang
L
,
Middha
S
,
Zheng
T
, et al
.
Identification of targetable kinase alterations in patients with colorectal carcinoma that are preferentially associated with wild-type RAS/RAF
.
Mol Cancer Res
2016
;
14
:
296
301
.
12.
Schlam
I
,
Swain
SM
.
HER2-positive breast cancer and tyrosine kinase inhibitors: the time is now
.
NPJ Breast Cancer
2021
;
7
:
56
.
13.
Gaibar
M
,
Beltrán
L
,
Romero-Lorca
A
,
Fernández-Santander
A
,
Novillo
A
.
Somatic mutations in HER2 and implications for current treatment paradigms in HER2-positive breast cancer
.
J Oncol
2020
;
2020
:
6375956
.
14.
Vaghi
C
,
Mauri
G
,
Agostara
AG
,
Patelli
G
,
Pizzutilo
EG
,
Nakamura
Y
, et al
.
The predictive role of ERBB2 point mutations in metastatic colorectal cancer: a systematic review
.
Cancer Treat Rev
2023
;
112
:
102488
.
15.
Kavuri
SM
,
Jain
N
,
Galimi
F
,
Cottino
F
,
Leto
SM
,
Migliardi
G
, et al
.
HER2 activating mutations are targets for colorectal cancer treatment
.
Cancer Discov
2015
;
5
:
832
41
.
16.
Nagano
M
,
Kohsaka
S
,
Ueno
T
,
Kojima
S
,
Saka
K
,
Iwase
H
, et al
.
High-throughput functional evaluation of variants of unknown significance in ERBB2
.
Clin Cancer Res
2018
;
24
:
5112
22
.
17.
Bekaii-Saab
T
,
Williams
N
,
Plass
C
,
Calero
MV
,
Eng
C
.
A novel mutation in the tyrosine kinase domain of ERBB2 in hepatocellular carcinoma
.
BMC Cancer
2006
;
6
:
278
.
18.
Raghav
KPS
,
Guthrie
KA
,
Kopetz
S
,
Tan
BR
,
Denlinger
CS
,
Fakih
M
, et al
.
A randomized phase 2 study of trastuzumab and pertuzumab (TP) compared to cetuximab and irinotecan (CETIRI) in advanced/metastatic colorectal cancer (mCRC) with HER2 amplification: SWOG S1613
.
J Clin Oncol
2023
;
41
:
140
.
19.
Peterson
S
,
Rosler
R
,
Klucher
K
.
Abstract 4222: Tucatinib, a selective small molecule HER2 inibitor, is active in HER2 mutant driven tumors
.
Cancer Res
2020
;
80
:
4222
.
20.
Guan
Y
,
Wang
Y
,
Li
H
,
Meng
J
,
You
X
,
Zhu
X
, et al
.
Molecular and clinicopathological characteristics of ERBB2 gene fusions in 32,131 Chinese patients with solid tumors
.
Front Oncol
2022
;
12
:
986674
.
21.
Bekaii-Saab
TS
,
Van Cutsem
E
,
Tabernero
J
,
Siena
S
,
Yoshino
T
,
Nakamura
Y
, et al
.
MOUNTAINEER-03: Phase 3 study of tucatinib, trastuzumab, and mFOLFOX6 as first-line treatment in HER2+ metastatic colorectal cancer—Trial in progress
.
J Clin Oncol
2023
;
41
:
TPS261
.
22.
Susanti
S
,
Fadhil
W
,
Murtaza
S
,
Hassall
JC
,
Ebili
HO
,
Oniscu
A
, et al
.
Positive association of PIK3CA mutation with KRAS mutation but not BRAF mutation in colorectal cancer suggests co-selection is gene specific but not pathway specific
.
J Clin Pathol
2019
;
72
:
263
.
23.
Ughetto
S
,
Migliore
C
,
Pietrantonio
F
,
Apicella
M
,
Petrelli
A
,
D'Errico
L
, et al
.
Personalized therapeutic strategies in HER2-driven gastric cancer
.
Gastric Cancer
2021
;
24
:
897
912
.
24.
Siena
S
,
Raghav
K
,
Masuishi
T
,
Yamaguchi
K
,
Nishina
T
,
Elez
E
, et al
.
386O Exploratory biomarker analysis of DESTINY-CRC01, a phase II, multicenter, open-label study of trastuzumab deruxtecan (T-DXd, DS-8201) in patients (pts) with HER2-expressing metastatic colorectal cancer (mCRC)
.
Ann Oncol
2021
;
32
:
S532
.
25.
Pietrantonio
F
,
Manca
P
,
Bellomo
SE
,
Corso
S
,
Raimondi
A
,
Berrino
E
, et al
.
HER2 copy number and resistance mechanisms in patients with HER2-positive advanced gastric cancer receiving initial trastuzumab-based therapy in JACOB trial
.
Clin Cancer Res
2023
;
29
:
571
80
.
26.
Woolston
A
,
Khan
K
,
Spain
G
,
Barber
LJ
,
Griffiths
B
,
Gonzalez-Exposito
R
, et al
.
Genomic and transcriptomic determinants of therapy resistance and immune landscape evolution during anti-EGFR treatment in colorectal cancer
.
Cancer Cell
2019
;
36
:
35
50
.
This open access article is distributed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) license.