Personalised cancer management—giving patients optimum treatment according to their individual circumstances (including genetics) and the molecular characteristics of their tumours—is a key theme of the European Society for Medical Oncology (ESMO) in 2013. Indeed, it is a key theme for oncologists in general, and in all aspects of medicine. Integrating research and innovation directed towards personalised care is also an objective of the European Union's Horizon 2020 science funding programme .
In cancer, personalised medicine is particularly crucial. In no other specialty are the margins between benefit and toxicity so small and the differences between patients so large. Given the narrow therapeutic index of the drugs we use, individual risk factors for toxicity (notably comorbidities and the extent and nature of prior treatments) are crucial in deciding the appropriateness of different interventions. This is evident, for example, with the risk of cardiotoxicity accompanying the use of trastuzumab .
We have to deal with profound differences between tumours arising in different organs but also with the potentially bewildering heterogeneity within tumour types. We need to take into account not only the clinical characteristics and histology traditionally used to guide treatment, but also the molecular abnormalities identified by increasingly sophisticated diagnostics. Nowhere is this more evident than in breast cancer, where choice of adjuvant chemotherapy in the individual patient can now be guided by a recurrence risk score based on the assay of 21 genes .
The recent ESMO symposium ‘Signalling Pathways in Cancer—targeting the HER/EGFR family’ (1–2 March 2013) discussed the implications for personalised medicine of the wealth of clinically significant information now becoming available from breast, lung and colorectal cancers (CRCs). While the implications for treatment are tumour-specific, the over-arching role of this signalling system provided a common theme.
achievements and lessons learned from HER2 targeting in breast cancer
Breast cancer has seen the greatest progress in targeting treatment to specific molecular abnormalities, particularly in the 20% of women whose tumours are driven by overexpression of HER2 . Indeed, we have known about the pathway for long enough to have clinical experience of the effects of different strategies, particularly in relation to blockade of the pathway at different levels. However, even 15 years after approval of the ground-breaking HER2 monoclonal antibody trastuzumab, we still grapple with the problem of heterogeneity within subgroups. In a sense, the more we learn, the less we know.
In part, this relates to predicting response to conventional chemotherapy, especially in the adjuvant setting, where the relatively small reduction achieved in absolute risk means that we are especially obliged to restrict toxic treatment to those most likely to benefit. The polygenic nature of drug resistance and inter- and intra-tumoural heterogeneity mean that identifying the optimal adjuvant cytotoxic regimen for individual patients remains a considerable challenge . For example, despite considerable effort, it has not proved possible to demonstrate that anthracyclines (as opposed to CMF) should be used only in women whose tumours amplify HER2 or have abnormalities in TOP2A .
There are also considerable problems caused by the fact that around half of HER2-positive tumours do not respond to anti-HER2 drugs. In the neoadjuvant setting, the high pathological complete response (pCR) rates achieved by dual HER2 blockade are mostly in ER-negative tumours. This illustrates the possibility of identifying tumours that are ‘truly addicted’ to HER2 signalling. Such efforts are crucial now that we are moving into the second and third generation of anti-HER2 drugs. We cannot afford to continue adding expensive agents to trastuzumab without biomarkers predictive of response or resistance to the anti-HER2 approach. The neoadjuvant setting may be very helpful in testing not only the potential efficacy of new agents, but also the predictive power of biomarkers.
Follow-up results of NOAH, a randomised, phase III trial, evaluating neoadjuvant chemotherapy with trastuzumab followed by adjuvant trastuzumab versus chemotherapy alone in patients with HER2-positive locally advanced breast cancer have been presented recently at 2013 Annual Meeting of the American Society of Clinical Oncology (ASCO). The updated analysis confirms the significant event-free survival benefit observed in the primary analysis of the study, and shows a strong trend towards improved overall survival (OS) with the addition of trastuzumab to chemotherapy. The authors pointed out that the pCR rate may be considered as a possible surrogate end point for long-term benefit, and the Food and Drug Administration is now ready to consider neoadjuvant trials for accelerated new drug approval .
In the NeoAdjuvant Lapatinib and/or Trastuzumab Treatment Optimization (NeoALTTO) trial, HER2-positive breast cancer patients received neoadjuvant lapatinib, trastuzumab or both (followed by paclitaxel, Taxol). Dual HER2 blockade almost doubled the rate of pCR, but not the proportion of women having breast-conserving surgery (BCS). Women with ER-negative tumours, with larger tumours and with multicentric disease, were less likely to have BCS . Surprisingly, radiological response did not influence the likelihood of BCS: does this show the absence of a personalised approach, leading to a large number of women having more radical surgery than they require? There is also evidence of a personalised approach, but one which does not seem to be based on the evidence: women with ER-negative tumours were less likely than those with ER-positive tumours to be considered for BCS, despite the fact that such tumours respond well to neoadjuvant chemotherapy and despite the absence of any data suggesting that BCS is inferior to mastectomy in this subgroup.
The impact of neoadjuvant chemotherapy plus HER2-targeting on breast conservation rates was studied prospectively in the CALGB 40601 trial. Surgical results presented at 2013 ASCO Annual Meeting show a nearly 50% conversion rate from BCS-ineligible to BCS-eligible patients with HER2-positive breast cancer. Furthermore, neoadjuvant chemotherapy combined with targeted anti-HER2 treatment permitted potential BCS in ∼80% of properly selected patients .
In advanced breast cancer, anti-HER2 therapy should be first line if OS is to be improved. In this setting, vinorelbine is a valid alternative to a taxane when coupled with trastuzumab . Preclinical data suggested that adding platinum to a taxane and then adding trastuzumab would be beneficial, but this has not been demonstrated in a randomized, clinical trial. The improved OS seen in the CLEOPATRA trial is strongly persuasive for the strategy of adding pertuzumab to trastuzumab plus docetaxel (Taxotere) in first-line . However, many patients had not been treated with adjuvant trastuzumab, and the outcome might have been different if they had had prior exposure to this agent. The Canadian trial in first-line HER2-positive metastatic breast cancer of a taxane plus trastuzumab or lapatinib showed longer progression-free survival (PFS) with trastuzumab . Interestingly, there is evidence of benefit from dual HER2 blockade in heavily pre-treated patients since trastuzumab plus lapatinib increased survival compared with lapatinib alone . The benefit of continuing anti-HER2 therapy beyond first progression is plausible but not proven. In patients where there is discordance in HER2 status between primary and secondary tumours, the consensus is to treat with an anti-HER2 agent. Tumours that do not amplify the HER2 gene but have mutations of the HER2 receptor probably account for no more than 2% of breast cancers, but may be a subgroup particularly responsive to the new irreversible anti-HER2 tyrosine kinase inhibitors .
Up to 40% of HER2-positive metastatic breast cancer patients develop brain metastases. According to the CEREBEL trial, the use of lapatinib plus capecitabine rather than trastuzumab plus capecitabine does not reduce this risk despite the fact that, as a small molecule, lapatinib should more easily pass the blood–brain barrier . Overall, lapatinib may have a relatively modest role compared with monoclonal antibodies, with its possible best use being in combination with trastuzumab. In the subgroup of patients with minimally symptomatic brain metastases, it may be considered in combination with capecitabine as a means of delaying whole-brain radiotherapy .
In the adjuvant setting, despite the fact that >50% of patients in the observation arm had been offered trastuzumab, the HERA study at 8 years still shows a strong survival benefit in patients initially randomised to the anti-HER2 antibody. Two years of trastuzumab is no better than 1 year, but it is not clear whether 6 months is sufficient. In the US NCCTG trial, the trend in disease-free survival favoured concomitant over sequential administration of trastuzumab with chemotherapy . Even though this difference did not reach statistical significance, concomitant administration is now preferred, except where there is particular risk of cardiotoxicity. The BCIRG006 study was not powered to compare anthracycline with non-anthracycline-based chemotherapy in combination with trastuzumab . With the current state of knowledge, patients without cardiac risk factors who are node positive should not be denied adjuvant anthracycline. In early HER2-positive breast cancer, dual HER2 blockade is likely to become the gold standard adjuvant treatment.
progress in EGFR signalling blockade and capturing the mechanisms of secondary resistance in CRC
For the management of colon and rectal cancers, current ESMO guidelines recommend a personalised approach to decision making based on factors related to the patient (including the ability to withstand major surgery and intensive chemotherapy), to the aims of treatment and to tumour biology .
In the adjuvant colon cancer setting, no biomarkers reliably predict particular benefit from adjuvant therapy. Neither do they help in the selection of specific agents. In advanced CRC, patients with K-ras-mutated tumours do not benefit from anti-EGFR (epidermal growth factor receptor) antibodies such as cetuximab and panitumumab. In such patients who are suited to intensive therapy, the recommended first-line combination is FOLFOX or XELOX plus bevacizumab. As yet, no markers predict subgroups of patients with K-ras-mutated tumours who are likely to gain particular benefit from anti-VEGF therapy .
In patients with wild-type K-ras, combining EGFR antibodies with chemotherapy is capable of converting initially non-resectable disease into disease amenable to surgery. In the pivotal study of single-agent cetuximab in CRC patients refractory to chemotherapy, the 0.77 hazard ratio for death (relative to best supportive care) in the overall population was further reduced to 0.55 when the analysis was confined to patients with wild-type K-ras tumours . The European data show that enriching a first-line, chemotherapy-refractory population for K-ras wild-type increases the objective response rate to cetuximab plus FOLFIRI from 24% to 36%. However, not all patients with wild-type K-ras benefit from anti-EGFR therapy (now typically FOLFIRI plus cetuximab). Moreover, the overall picture in the first-line setting is mixed, especially in relation to the addition of cetuximab to oxaliplatin-based combinations. NORDIC-VII showed no OS benefit, even in wild-type K-ras patients . In the UK Medical Research Council's COIN study, the addition of cetuximab to standard therapy increased the response rate but again had no effect on OS .
Downstream effectors of EGFR, including mutations in BRAF and PI3K pathways, have been evaluated. BRAF mutations cause decreased efficacy to cetuximab and panitumumab, and when considered with K-ras status, the combined molecular assessment can identify up to 55% of CRC patients who will not respond. The existence of the BRAF mutation does not confer that a single agent targeting this mutation will be effective. Combination therapies to overcome primary and secondary resistance are under investigation. More refined prediction of response will probably require analysis not of single-gene mutations but of gene signatures. Although progress is being made, this is still a distant goal, and the possible relevance of BRAF and PI3KCA mutations as a means of further sub-classifying CRC patients for targeted treatment selection remains investigational.
The mechanisms responsible for secondary EGFR resistance have been described as ‘myriad’. Their investigation is made difficult by a natural reluctance to re-biopsy patients but identifying those at risk may be helped by studies of circulating DNA to detect the emergence during treatment of K-ras mutations . Molecular alterations (in most instances point mutations) of K-ras are causally associated with the onset of acquired resistance to anti-EGFR treatment; secondary K-ras mutations are responsible for acquired resistance in ∼50% of the CRC patients who initially respond to cetuximab or panitumumab. Furthermore K-ras mutant clones can be detected non-invasively in blood months before radiographical progression . Some data suggest early initiation of an MEK inhibitor as a rational strategy for delaying or reversing drug resistance.
The results from a study of the molecular bases of relapse in patients with metastatic CRC who do not develop K-ras mutations during the course of anti-EGFR therapy have been recently presented at the 2013 ASCO Annual Meeting. Next generation sequencing was applied to tumour biopsies to identify genetic alterations associated with relapse to cetuximab and panitumumab. The molecular analyses revealed a high level of amplification of the MET proto-oncogene. To functionally evaluate the role of MET amplification in resistance to anti-EGFR antibody therapies, the researchers exploited patient-derived CRC xenografts (‘xenopatients’). They found that xenopatients established from MET-amplified tumours were completely refractory to cetuximab but showed sensitivity to the MET inhibitor crizotinib . The emergence of MET amplification in circulating, cell-free, DNA may be used to select patients most likely to benefit from anti-MET therapies.
HER2 amplification present in a subgroup of CRC patients represents a potential target. The clinical value of rechallenging patients with anti-EGFR antibodies needs to be demonstrated, as does the potential of novel inhibitors of downstream signalling in the EGFR pathway. When given with oral chemotherapy, the combination of anti-EGFR and anti-VEGF antibodies has deleterious effects compared with the use of the anti-VEGF approach alone . However, there may be some subgroups—including highly symptomatic patients in whom response is desirable—who could benefit.
mapping the mutational landscape and dissecting the EGFR/HER pathway in targeted treatment of lung cancer
In non-small-cell lung cancer (NSCLC) patients with advanced disease, anti-EGFR agents were the first breakthrough in tailoring treatment according to the molecular characteristics of the tumour . The initial studies were randomized trials of the first-generation EGFR inhibitors given as second-line versus placebo. One study with erlotinib was positive and one study with gefitinib was so-called negative; both studies showed a marginal improvement of survival or a significant improvement of survival compared with placebo given as second- or third-line treatment. The real change came at ESMO 2008 with the presentation of the study carried out in what we now know as an enriched population of patients with EGFR mutations. This study compared a first-line gefitinib treatment with a standard chemotherapy and it showed superiority in PFS and response with gefitinib treatment. Since this landmark trial, there have been a number of studies which confirmed the better response and superior PFS with first-line targeted treatment versus chemotherapy.
Testing for EGFR mutation (and now also ALK rearrangement) before using tyrosine kinase inhibitors (TKIs) that target these genetic abnormalities is now standard, showing how far personalised medicine has advanced in the management of adenocarcinoma of the lung . In patients who have an EGFR mutation, the question of whether anti-EGFR agents should be given first- or second-line remains open, since survival seems similar. Anecdotal evidence suggests that, in contrast to chemotherapy, anti-EGFR drugs should not be discontinued when a patient with a previously responsive EGFR tumour shows signs of progression.
Among patients with an activating EGFR mutation, a subgroup also has a T790M mutation which confers relative resistance to erlotinib and gefitinib. Ongoing trials should elucidate how best to treat these patients Activating mutations in the tyrosine kinase domain of HER2 have been described in a subset of lung adenocarcinomas as mutually exclusive with EGFR and K-ras mutations. Arcila et al. reported that HER2 mutation was significantly associated with NSCLC patients who were never smokers, but did not associate with sex, race or disease stage. This group reported that HER2 mutations identify a distinct subset of lung adenocarcinomas. Given the high prevalence of lung cancer worldwide and the availability of standard and investigational therapies targeting HER2, the group advocated that routine clinical genotyping of lung adenocarcinoma should include HER2 . However, no association between HER2 mutation and HER2 overexpression was shown in their study and whether patients with these mutations truly benefit from anti-HER2 strategies remains to be demonstrated.
In a recent molecular profile analysis of 2271 cases of NSCLC, EGFR was mutated in 12% and K-ras in 32% of cases; HER2 gene amplification was confirmed as a rare event in NSCLC (4%); coexistence of HER2 gene amplification and EGFR mutation was identified in 3 cases, while K-ras was mutated in 7 HER2-amplified cases; double EGFR mutations were however found in only 2 cases. NSCLC with HER2 amplification were frequently (39%) associated with K-ras activating mutation. A rare A859T mutation was found in one case and was associated with HER2 gene amplification; this mutation was previously associated with resistance to TKIs. This molecular insight in large samples of NSCLC cases was presented at the 2013 European Multidisciplinary Conference in Thoracic Oncology . Since earlier studies have suggested that HER2 amplification may cause resistance to erlotinib and gefitnib, NSCLC patients with HER2 amplification and activating EGFR mutation may respond better to afatinib, which inhibits both HER2 and EGFR activities.
Besides this recent recognition that HER2 amplification may be a mechanism of acquired resistance to EGFR-TKI that occurs in a subset of tumours lacking the acquired, somatic EGFRT790M mutation, the amount of research dedicated to HER2 in NSCLC is increasing. HER2-mutated NSCLC represents a small distinct subgroup of oncogene addicted cancers with specific demographics and potential outcomes . The results of the largest clinical analysis to date in patients with NSCLC and HER2 mutations, despite limitations of studying data retrospectively, provide important insights into HER2-driven NSCLC. Furthermore, this study reinforced the importance of screening for HER2 mutations in lung adenocarcinomas and suggests the potential efficacy of HER2-targeted drugs in this population. Clinical trials of trastuzumab, pertuzumab and afatinib are ongoing.
Lung cancer is a very good example of showing potentials, but also challenges accompanied with molecular classification of tumours. If you have a patient with a disease seen in 2% of all patients and then resistance mechanisms develop in 10 different ones, how rare are those patient subgroups that you have to identify to determine the value of a specific treatment? Personalised medicine not only means the choice of treatment tailored to the individual patient, but also reflects the need for considering how investigative and diagnostic strategies must also be planned according to individual tumour characteristics. To allow the extended diagnostic and predictive examination, tissue sampling should be maximised whenever feasible and deemed clinically safe, reducing the need for re-biopsy for additional studies. In addition tissue handling, processing and sectioning should be optimised .
Complex diagnostic algorithms are emerging and personalised medicine is now a reality for advanced NSCLC patients on the basis of routine screening for EGFR mutation and ALK gene fusion assessment. At the 2013 ASCO Annual Meeting, we have heard that the French National Cancer Institute (INCa) also decided to fund the routine assessment of four additional biomarkers (HER2, KRAS, BRAF, PI3KCA). Starting in April 2012, these biomarkers were prospectively collected in a database and until January 2013 10 000 biomarker analyses were collected. The physicians who ordered each of these biomarker analyses were asked to regularly complete epidemiological, clinical and therapeutic data for each corresponding patient. The molecular profile was characterised by 9.4% EGFR (including 0.8% EGFR resistant), 0.9% HER2, 26.9% K-ras, 1.6% BRAF, 2.6% PI3KCA mutated tumours and 4.0% EML4-ALK fusion genes. Double mutations were seen in 0.9% of the tumours . At the time of analysis, data on treatment were available for 18.6% of patients among whom 56.9% of patients received a treatment according to their molecular profile (labeled drugs or a treatment within bio-guided clinical trials). It is the largest ever conducted biomolecular study in patients with advanced NSCLC and provides encouraging data on the value of a nationwide biomarker screening policy.
progressing towards personalised cancer care
In the overall context, our step-by-step exploration of the HER/EGFR pathway, with all its twists, turns and occasional blind alleys, is a model of the journey towards personalised cancer care based on the analysis of abnormal signalling; and the insights derived apply across several tumour types. Clearly, that journey has only just begun. At the moment, much of our further effort is still fragmented. Turning the dream of personalised oncology treatment into reality will require a collaborative endeavour that matches the scale of the problem. The primary component is research, which must be international in scope, and as unencumbered as possible by bureaucratic regulation. But it also requires educational initiatives to disseminate and interpret from a critical perspective the results of research to physicians and to patients. In both research and education, we all have a part to play.
MP declares the following conflicts of interest: Board member: PharmaMar; Consultant (honoraria): Amgen, Astellas, AstraZeneca, Bayer, Invivis, MSD, Novartis, Pfizer, Roche-Genentech, Sanofi-Aventis, Symphogen, Synthon, Verastem; Research grants to my Institute: most companies; Speakers bureau/stock ownership : none.