Improving disease control in advanced colorectal cancer: Panitumumab and cetuximab

3.1. Preclinical development 

Panitumumab was generated using XenoMouse strains, engineered to be deficient in mouse antibody production and to contain integrated fragments from the human heavy and Kappa light chain loci [28]. Panitumumab exhibited a high affinity to the EGFR, higher than that reported for other anti-EGFR Moabs [29], as well as the ability to neutralise both EGF and TGF-alfa binding to EGFR-expressing human carcinoma cell lines [30].

Besides EGFR signalling blockade, other mechanisms may account for panitumumab effect on different tumour cell lines, including antiangiogenic effects through inhibition of VEGF and IL-8 production and matrix metalloproteinases inhibition [31], [32].

In vivo, panitumumab mediates therapeutic elimination of established tumours and shows synergies with chemotherapy in mediating tumour regression [33].

These observations provided the rationale for the design of a phase I multicentre, open-label, dose escalating clinical trial in 96 patients with a variety of EGFR-expressing human solid tumours. No infusion-related or serious adverse events were noted (a transient and dose dependent skin rash was the primary toxicity). No maximum tolerable dose (MTD) was reached. Pharmacokinetic (PK) exposure was similar between weekly 2.5mg/Kg, 6mg/Kg every 2 weeks and 9mg/Kg on a 3-weekly basis, with low inter and intra-patient variability [34]. Since preliminary evidence of antitumoural activity was suggested in aCRC [35], specific phase II trials with this compound were initiated in this condition.

3.2. Clinical trials with single-agent panitumumab 

The safety and efficacy of single-agent panitumumab has been recently summarised across five clinical studies involving 762 aCRC patients relapsed after oxaliplatin and/or irinotecan-based chemotherapy. This analysis that did not take into account patient's Kras status showed a disease control rate in the range of 29–44% and a median progression-free survival of 8–14 weeks [36]. In view of these consistent results, a single phase III, open-label, randomised, multinational study [37] was initiated (Table 1). 463 EGFR-expressing (at least 1+ membrane staining in >1% of tumour cells) aCRC patients were randomly assigned to best supportive care (BSC) with or without panitumumab, 6mg/kg i.v., every 2 weeks. The primary study endpoint was PFS. BSC patients who progressed were eligible to cross over to receive panitumumab on progression. All patients enrolled were refractory to standard chemotherapy regimens containing fluoropyrimidines and oxaliplatin or irinotecan. 100% and 37% of the patients had received 2 and 3 lines of previous treatment, respectively. The median PFS was 8 weeks in the panitumumab arm and 7.3 weeks in the BSC group, with 49% of the patients in the panitumumab group and 30% in the BSC group being alive and progression-free by week 8. Panitumumab reduced the risk of disease progression vs. BSC alone by almost half (hazard ratio=0.54, 95%CI: 0.44–0.66, p<0.0001). The mean PFS was longer in the panitumumab arm than in the BSC arm (13.8 vs. 8.5 weeks). In patients unselected for Kras status the ORR was 10% in the panitumumab group, with a median duration of response of 17.0 weeks, whereas no responses were observed in the BSC arm. Stable disease (SD) was observed in 27% and 10% in the panitumumab and BSC groups, respectively. No responders were identified in the panitumumab mutant Kras group, whereas wild-type Kras patients treated with panitumumab achieved a 17% ORR [38]. Among patients who crossed over to receive panitumumab after progression, ORR was 11% with an additional 33% of patients achieving SD [39]. All responders had wild-type Kras, for an overall response rate in this subgroup of patients of 22% [38]. Median progression-free survival was 9.4 weeks and median overall survival 6.3 months.

Table 1. Clinical trials with panitumumab in monotherapy or combination therapy. Number of patients (n); overall response rate (ORR); progression-free survival (PFS); overall survival (OS); best supportive care (BSC); weeks (w); months (m). Irinotecan, bolus 5-fluorouracil, and leucovorin (IFL). Irinotecan, infusional 5-fluorouracil and leucovorin (FOLFIRI).

	Treatment	Patients (n)	ORR (%)	PFS	OS (m)	Pre-treated	Phase
	Panitumumab vs. BSC [37]	231	10	8w	–	Yes	III
		232	0	7w	–
	Panitumumab after BSC [39]	176	11	9w	6	Yes	III cross over
	Panitumumab [52]	148	9	14w	9	Yes	II
	Panitumumab+IFL [41]	19	46	6m	17	No	II
	Panitumumab+FOLFIRI [41]	24	42	11m	23	No	II
	Panitumumab [53]	91	7	8w	–	Yes	II

On this basis, the U.S. FDA approved panitumumab (Vectibix®), for the treatment of patients affected by EGFR-expressing metastatic colorectal cancer (with >1% EGFR tumour cell membrane staining) after disease progression on fluoropyrimidine, oxaliplatin and irinotecan containing chemotherapy regimens [10]. In addition, the European Committee for Medicinal Products for Human Use also issued a positive opinion, approving panitumumab for use in the European Union (EU) for patients with Kras wild-type, chemotherapy refractory aCRC [38].

3.3. Panitumumab-based combinations 

The encouraging results seen with single-agent panitumumab in heavily pre-treated aCRC patients and the reported preclinical synergistic interactions [33] have prompted the design of several panitumumab-based combination trials.

Mature results of a two-part multicentre study with the combination of panitumumab and 5-fluorouracil (5-FU), leucovorin and irinotecan, either IFL or FOLFIRI [38], in aCRC patients are now available. For both chemotherapy regimens, ORR was more than 40% and disease control rate almost reached 80%. Despite previously reported lack of significant PK interactions between panitumumab, irinotecan and SN-38 when the combination of panitumumab and IFL was first analysed [40] these data suggest that panitumumab may have better tolerability when combined with FOLFIRI than with IFL. In fact, grades 3/4 diarrhoea occurred in 11 patients (58%) treated with IFL-panitumumab compared with only 6 patients (25%) in the FOLFIRI-panitumumab regimen [41].

Ongoing randomised phase III trials are evaluating the addition of panitumumab to either FOLFIRI [42] in pre-treated aCRC patients or to FOLFOX (PRIME study) [43] as frontline therapy. Preliminary data suggest no pharmacokinetic/pharmacodynamic interactions and an acceptable safety profile, as such both trials continue without modifications. Enrolment was anticipated to be completed by the end of 2008 and results are eagerly awaited. Other ongoing clinical trials with panitumumab are listed on-line at (http://clinicaltrials.gov/ct2/results?term=panitumumab+colorectal+cancer).

3.4. Safety profile 

The most commonly reported adverse events in the available panitumumab studies are skin rash, hypomagnesemia, paronychia, fatigue, abdominal pain, nausea, and diarrhoea. The most serious adverse events, albeit rarely seen include pulmonary fibrosis, severe dermatological toxicity complicated by infectious sequelae and septic death, infusion reactions, abdominal pain, hypomagnesemia, nausea, vomiting, and constipation [44].

Skin related adverse events are seen in the vast majority of aCRC patients treated with EGFR inhibitors, although they result in discontinuation of therapy in only 2% of patients. The incidence of grade 3 skin toxicity is in the range of 15%. Most frequently reported dermatological manifestations include acneiform dermatitis, erythema, pruritus, rash, skin exfoliation, telangiectasia, trichomegaly and hirsuitism [45], [46]. The most commonly occurring eye disorders are conjunctivitis (4%), ocular hyperemia (3%), and increased lacrimation (2%). Nail toxicities are identified in around 30% of the patients treated with panitumumab and include paronychia (24%), nail disorder (9%), onchorrhexis, nail bed infection, nail bed inflammation, nail discoloration, nail discomfort, and onycholysis (all 1%) [10], [37].

Gastrointestinal toxicities are also commonly reported and include nausea, diarrhoea and anorexia. Most of them are mild, being grade 3 in only 2% of the treated patients [46].

Infusion reactions were noted in 4% of patients and in 1% reactions were graded as severe (National Cancer Institute Common toxicity Criteria [NCI-CTC] grades 3–4). Across all clinical studies, severe infusion reactions, including anaphylactic reaction, bronchospasm, fever, chills, and hypotension, occurred in less than 1% of patients. No infusion reaction was fatal [46].

Hypomagnesemia [47] appears in approximately 40% of patients treated with panitumumab. Hypomagnesemia often occurs after the 6–8 weeks and may be associated with hypocalcemia. NCI-CTC grade 3 or 4 hypomagnesemia requiring oral or i.v. electrolyte repletion occurred in 2% of panitumumab monotherapy-treated patients [48].

4.1. Skin rash 

A peculiar toxic effect of panitumumab, and other EGFR inhibitors, is a papulopustular skin rash, generally on face and upper torso, which is thought to be mechanism- and dose-related [49]. Several findings suggest that there is a correlation between intensity of skin rash and response and survival [5]. With respect to panitumumab, this correlation is particularly striking in a recently reported pooled analysis from 5 panitumumab clinical trials involving 612 aCRC patients. In this large combined analysis, severity of skin rash (grades 2–4 vs. grades 0–1) was significantly correlated with an increased RR (12.6% vs. 3.3%), longer median PFS (13.1 weeks vs. 8.0 weeks) and median OS (8.5 months vs. 4.5 months) [50].

The mechanism underlying the correlation between skin toxicity and tumour response is currently unclear, although some groups hypothesised that the rash is a surrogate indicator of an adequate degree of receptor saturation by anti-EGFR therapy. Pharmacokinetic data have shown that EGFR-mediated clearance of panitumumab was saturated at a dose of 2mg/kg. A 100% skin rash incidence was achieved with increasing doses to 2.5mg/kg, suggesting full receptor occupancy at this dose level [51].

Although skin rash may be a pharmacodynamic marker of drug action its predictive value and potential as a surrogate marker for response to EGFR-targeted agents depend on the correlation of HER-kinase signalling in paired skin and tumour tissue, an issue that requires further investigation.

4.2. HER family expression 

To date, EGFR positivity by immunohistochemistry (IHC) analysis on tumour specimen has been mandatory for clinical use of panitumumab, as well as for eligibility in most clinical trials with this drug. In fact, a >1% EGFR tumour cell membrane staining was established as a necessary condition for panitumumab US FDA approval for this indication.

This biomarker however, has not demonstrated a clear role in predicting clinical response of aCRC to EGFR TKIs or Moabs [18], [29]. All the reported trials with panitumumab have shown that tumour response does not correlate with the degree of EGFR expression. In fact, objective responses have been observed in patients with low or negative EGFR levels. Hetcht et al. reported a phase II open-label, multicentre study in 148 EGFR-positive aCRC patients who had progressed to fluoropyrimidine, irinotecan and/or oxaliplatin-based regimens (Table 1) [52]. All patients received weekly panitumumab 2.5mg/kg, 8 of out of 9 weeks until disease progression. Interestingly, a 5% RR was seen in patients with low or non-EGFR staining. Mitchell et al. confirmed these results in a single-agent panitumumab trial in 91 pre-treated aCRC patients with negative or low EGFR expression by IHC. Disease control rate (DCR) was achieved in 36–42% of patients with a median duration of response of 20–22 weeks and a median PFS of 8 weeks, as shown in Table 1 [53]. Several factors might explain this apparent discrepancy, such as low sensitivity of IHC analysis, cytological heterogeneity of CRC and differential EGFR expression in primary and metastatic tumour tissue [54]. Finally, in some cancer cells, regardless the EGFR expression level, critical downstream signals may be activated via other receptors, or by other pathways [55]. Thus, a current important challenge in EGFR-targeted therapy is to better identify those tumours in which growth is most dependent on EGFR signalling.

In addition to EGFR, HER2 expression has been related to responsiveness to EGFR-targeted therapy. Her-2/neu is known to heterodimerise preferentially with EGFR [56], and to potentiate mitogenic signalling by increasing EGFR ligand affinities [57], [58]. In human cancer cell lines, increased HER2 expression levels alone were associated with increased effectiveness of anti-EGFR therapy. Moreover, the presence of a combined EGFR and HER3 overexpression proved to be an even better predictor of response [59]. These findings have been confirmed in the clinical setting for lapatinib, a dual inhibitor of EGFR and HER2, since co-expression of pHER2 and pHER3, and not pEGFR, better correlated with clinical outcome [60]. Further research to confirm these data with other anti-EGFR therapies seems advisable.

4.3. Kras mutations 

Alternative predictive biomarkers to EGFR IHQ expression are being actively pursued. Different studies suggest that Kras gene mutations are present in 30–50% of colorectal adenocarcinomas [61] and these mutations would be responsible for a constitutive activation of Kras protein downstream EGFR.

The status of Kras has been proven to be a major predictor of panitumumab efficacy. In a study by Amado et al. [38] Kras status was analysed in 427 of 463 aCRC patients treated with panitumumab. Median PFS was 12.3 weeks in the wild-type (WT) Kras group compared to 7.4 weeks in Kras mutant (MT) patients (p<0.0001), and ORR was 17% and 0%, respectively. More recently, a pooled analysis from 4 clinical studies of panitumumab confirmed Kras as a valid predictive biomarker in selecting aCRC patients for panitumumab monotherapy. Kras genotype was identified using RT-PCR, from primary or metastatic tumour samples. ORR in patients with WT Kras samples was 13.7% whereas no responses were observed in patients with MT Kras samples. PFS and OS also favoured WT Kras group, with a HR of 0.42 (95%CI: 0.36–0.50) and 0.63 (95%CI: 0.53–0.74), respectively [62].

Although the presence of mutated Kras alleles is an independent predictive maker of anti-EGFR MoAb resistance it only accounts for 30–40% of non-responsive patients. Subsequently, efforts have focused on assessing whether the mutation status of the genes for other intracellular effectors (BRAF, PIK3CA) correlate with clinical response to anti-EGFR MoAbs. In this sense, in a recent retrospective study of 113 aCRC patients who received cetuximab or panitumumab, Kras and BRAF point mutations in exons 2 and 15 were analysed, respectively [63]. The data confirmed that the presence of Kras mutations correlated with both, a lack of response to anti-EGFR MoAbs therapy and a shorter PFS. Since Kras and BRAF mutations are mutually exclusive, the authors analysed the BRAF mutational status on WT Kras tumours. Notably, the presence of BRAF mutations was inversely associated with response to therapy. None of the WT Kras patients who experienced response to therapy displayed BRAF mutations. In addition, in the subgroup of WT Kras patients, the presence of BRAF mutations correlated with shorter PFS and OS. The authors propose a combined analysis of both, Kras and BRAF, to prospectively select aCRC patients eligible for EGFR-targeted MoAbs therapy. Nevertheless, these findings need to be further evaluated in a large, prospective phase III trial, given that the number of patients in each subgroup was relatively small.

Besides Kras and BRAF mutations, key mechanisms involved in resistance to anti-EGFR therapy appear to emerge from the constitutive activation of other EGFR downstream signalling phosphoprotein expression. By using Bioplex phosphoprotein array in aCRC patients treated with cetuximab or panitumumab, Cox multivariate analysis adjusted for age and gender found Kras and pMEK to be two independent prognostic makers [64]. These data highlight the importance of EGFR downstream signalling phosphoprotein expression as predictive biomarkers in this setting.

4.4. EGFR gene amplification 

It has been assumed that aCRC responsiveness to anti-EGFR MoAbs was linked to increased EGFR gene copy number (GCN) as measured by FISH. However, preliminary results have been inconclusive due to small sample size and treatment heterogeneity [65], [66]. The predictive role of EGFR GCN has recently been suggested in 64 aCRC patients treated with panitumumab therapy [67]. A mean EGFR GCN of less than 2.5/nucleus or less than 40% of tumour cells displaying chromosome 7 polysomy within the tumour correlated with significantly shorter PFS and OS. The most appropriate cut-off values and the reproducibility of FISH results should be confirmed in larger trials. In addition, a prospective validation to find out whether detection of EGFR GCN together with downstream signalling phosphoprotein mutations (Kras, BRAF) can be clinically exploited for optimisation of anti-EGFR MoAb therapy warrants further research.

4.5. EGFR polymorphisms 

EGFR polymorphisms have also been suggested as candidate predictors of clinical outcome in patients treated with anti-EGFR therapies. Carcereny et al. [68] analysed 84 aCRC patients who received cetuximab or panitumumab as second or third line therapy. They found single nucleotide polymorphisms (SNP) in codon 216 and 497 and a dinucleotide repeat polymorphism in intron 1. Interestingly enough, SNP 497 was significantly associated with a lower ORR and a shorter PFS and OS. Importantly, by multivariate analysis, pre-treatment CEA level, LDH and EGFR SNP 497 were identified as independent prognostic factors for PFS and OS. Prospective validation in a larger cohort of patients seems advisable.

5.1. Combination with other targeted therapies 

Although activation of EGFR may be an important signal for tumour cell growth and survival, other growth factor receptors may be crucial for the proliferation and dissemination of neoplastic cells [69]. This assumption leads to the so-called “combined molecular targeting approach” in which more than one class of inhibitors is applied simultaneously. There are expanding series of such approach being currently tested in clinical trials. Among them, the combined blockade of angiogenic pathway and EGFR signalling pathway is being actively investigated.

The link between EGFR signalling and angiogenesis has been clearly identified in preclinical models [70], [71]. Taking into account this rationale, as well as the promising results from the BOND-2 trial, a phase IIIB randomised, open-label clinical trial evaluating oxaliplatin- and irinotecan-based chemotherapy plus bevacizumab, with or without panitumumab, was initiated. Unexpectedly, preliminary data on this study showed an increased incidence of serious adverse events in the panitumumab arm as well as reduced median PFS and OS. Consequently, the Data Monitoring Committee recommended discontinuation of panitumumab [72]. The lack of benefit of adding panitumumab to standard chemotherapy plus bevacizumab reinforces the need for robust phase I/II-based clinical evidence before large phase III trials are conducted. In this sense, no data on the specific interaction between panitumumab and bevacizumab are currently available, nor on the possible PK/PD interactions with the use of two biological agents and full doses of chemotherapy. Another recently reported trial testing the combination of oxaliplatin, capecitabine and bevacizumab, with or without cetuximab also reported a detrimental effect on PFS with the double-antibody therapy, raising the possibility of a negative interaction between bevacizumab and anti-EGFR antibodies [73].

A more cautioned approach has been reported in a phase Ib, dose-finding study of panitumumab with FOLFIRI or FOLFOX plus AMG 706, an oral multikinase inhibitor with antiangiogenic activity that selectively targets VEGF, PDGF and Kit receptors. So far, described dose limiting toxicity (DLT) includes diarrhoea and fatigue. Preliminary data show that AMG 706 PK is comparable to data from monotherapy studies and no marked alterations have been observed in the PK profiles of irinotecan or its metabolites. According to the preliminary data reported, radiological responses have been observed in 50% of the patients [74].

On the other hand, recent evidence has involved the insulin-like growth factor-1 receptor (IGF-1R) signalling through the type II receptor tyrosine kinase (RTK) family member and the acquisition of resistance to various anti-EGFR therapies [75]. In preclinical models, blockade of IGF-1R results in both, a potent inhibition of the PI3K/Akt signalling pathway and additive effects with panitumumab [76]. In the clinical scenario, a phase Ib of panitumumab plus AMG-479, a fully human anti-IGF-1R antibody, appeared tolerable and has shown preliminary evidence of activity in refractory aCRC [77].

5.2. Role of panitumumab in cetuximab-pre-treated patients 

Panitumumab and cetuximab are the first anti-EGFR Moabs approved for the treatment of aCRC, showing both of them a similar safety and efficacy profile, when compared to BSC (Table 2, Table 3). By using flow cytometry and amino acid replacement scanning it was shown that both Moabs bind to comparable surface exposed amino acids (349, 355, 412 and 438) in domain III of EGFR [78]. Despite this structural homology some clinical differences have been observed. Given the fully human nature of panitumumab, a lack of cross reactivity with cetuximab regarding infusional reactions may be expected. However, this issue has not been prospectively explored. Retrospective data suggest however that panitumumab is not associated with hypersensitivity in patients who had prior grades 3–4 infusional reactions to cetuximab [79], [80]. In addition, a differential response to these two anti-EGFR Moabs has also been suggested. Furthermore, single cases of CEA responses and minor responses to panitumumab have been suggested in patients who had evidence of progression to cetuximab [81]. Although theses data are so far inconclusive, due to the crucial therapeutic implications of this finding a thorough prospective validation in larger series of patients is warranted.

Table 2. Grade 3 or higher toxicities with an incidence of ≥5%. Nausea/vomiting (N/V); best supportive care (BSC).

		Cetuximab+BSC [21]	Panitumumab+BSC [37]
	Fatigue	33%	4%
	Anorexia	8%	3%
	Constipation	4%	3%
	N/V	6%	2%
	Abdominal pain	13%	7%
	Rash	12%	7%
	Infusion reaction	4%	1%
	Hypomagnesemia	6%	3%
	Dyspnea	16%	5%
	Fatigue	33%	4%

Table 3. Efficacy data comparing cetuximab and panitumumab. Overall survival (OS); progression-free survival (PFS); overall response rate (ORR); wild-type (WT); mutant-type (MT); weeks (w); months (m).

		Cetuximab [90]	Panitumumab [37]
	Median OS
	WT	10m	8m
	MT	5m	5m
	Medias PFS
	WT	4m	12w
	MT	2m	7w
	ORR (%)
	WT	13	17
	MT	1	0