The maximum tolerated dose of 50 mCi by the intra-arterial arm in the present study is difficult to compare with the results of other studies, as we did not dose patients based on their body surface area. Being a phase I exploratory study, we used pre-specified dose levels in terms of mCi. However, on post-hoc review of the dose based on body surface area, patients in whom dose limiting toxicity was seen i.e. those administered 50 mCi by the intra-arterial route received a median dose of 27.5 mCi/m2. This is lower than the maximum tolerated dose of 60 mCi/m2 reported by Behr et al , as well as the 40 mCi/m2 found by Hajjar et al .
Surprisingly, despite the loco- regional nature of delivery of KAb201, systemic toxicity occurred. Directly labelled monoclonal antibodies are known to have a relatively low level of uptake in solid tumours, and additionally hampered by heterogeneous deposition of the antibody as well as radiation doses in tumour tissue . The unbound radiolabelled KAb 201 circulating systemically is likely to be responsible for the toxicity observed.
By the intravenous route, a larger dose of radiation, up to 75 mCi at close of trial, equating to a median of 45 mCi/m2, could be delivered without causing dose limiting toxicity. The haematological toxicity seen is in keeping with previous reports, which also reported myelotoxicity to be the main dose limiting factor [16, 31, 32]. Dosimetry based planning of treatment and pretargeting may minimise this problem . Since the aim of this phase I trial was to determine the MTD for KAb 201 with I131 and the therapy doses were planned as per protocol, we did not use formal dosimetry.
A poor correlation between the toxicity grade and administered radioactive dose has been reported, leading to the conclusion that factors other than the total administered activity or the bone marrow dose are important . The incidence of systemic toxicity in the intra-arterial arm implies that despite the loco regional delivery, there is spill over into the systemic circulation, supported by the finding of similar volume of distribution in both arms. The earlier occurrence of haematological toxicity in the intra-arterial arm could be linked to the slower rate of elimination seen in this arm, compared to the intravenous arm.
The similar pharmacokinetics of KAb201 and I131 implies it was appropriate to combine these two agents, as their decline runs in parallel. In view of the fact that the levels reach near undetectable levels by 6–8 weeks in most patients, this time point, rather than the 3 months chosen at start of study, may be more appropriate for either a repeat dose or commencement of palliative treatment off-trial.
The antigenic response seen in the majority of patients to both the sheep and chimaeric component of the antibody limits the possibility of repeat dosing, as this may either lead to hypersensitivity reactions or complexing with circulating antibody, making it difficult or impossible to maintain effective therapeutic levels . Repeat dosing using high affinity humanised monoclonal anti-CEA antibody have been reported [16, 33].
Similar to our antigenicity result, Ritter et al, despite using a humanised murine monoclonal antibody, detected human anti-human antibodies (HAHA) in 63% of patients treated with repeat dosing . They suggest that monitoring the type of antibody helped them single out patients at risk for transfusion-related adverse events, as those who developed type I HAHA (characterised by early onset, with levels peaking after 2 weeks and declining thereafter) did not develop infusion-related adverse events, unlike patients with type II antibodies (onset delayed, with levels increasing following repeat dosing).
The overall response rate of 6% is similar to that reported using single agent gemcitabine by Burris et al (5.4%) , as well as in two other large studies by Cunningham et al (7%)  and Moore et al (9%) .
The effectiveness of monoclonal antibodies has been limited by low quantitative delivery to tumours, poor diffusion from vasculature into tumour and biodistribution to normal organs . Although tumour vessels have attributes that favour movement of molecules across the vessels such as wide inter-endothelial junctions, large number of fenestrae and discontinuous or absent basement membrane, these are offset by the high interstitial pressure and low microvascular pressure that may retard extravasation of molecules, particularly into large tumours . Other factors limiting efficacy may be inherent radio-resistance and heterogenous expression of CEA .
Our study's median overall survival of 5.2 months is comparable to that achieved by single agent gemcitabine in several trials of 4.0–5.2 months [6, 37, 38], and somewhat inferior to the median survival of 6.0 to 7.3 months in the single agent gemcitabine arm in some others [34, 39–41]). Survival with I131 KAb201 may be boosted by combination with chemotherapy, which may also help radiosensitize the tumour. The chemotherapy suggested, based on a recent meta-analysis, is gemcitabine based combination chemotherapy . Alternatively, to avoid further increasing marrow toxicity, combination with erlotinib may help improve its effectiveness.