Summary and interpretation of findings
The dose of 5-fluorouracil-containing regimens given to cancer patients is widely based on the patient’s body surface area, but about 40 %–50 % of patients receiving 5-fluorouracil in this way may be under-dosed. Plasma 5-fluorouracil estimation in conjunction with dose adjustment algorithms might achieve more appropriate 5-fluorouracil dosing. We systematically reviewed the evidence on the clinical effectiveness and safety of pharmacokinetic dosing relative to dosing based on body surface area.
Although we identified 19 publications investigating clinical outcomes for pharmacokinetic 5-fluorouracil dose adjustment regimens only three studies compared PKA versus BSA, and only one of these was randomised. It is clear from the three comparative studies that there is an apparent advantage in PK monitoring for both progression-free survival and overall survival. Except for hand and foot syndrome, the particularly frequent and undesirable adverse events associated with 5-fluoruracil administration appear to be reduced and or delayed in these comparative studies so that taken in the round, these studies indicate that a PK dosing strategy is unlikely to be harmful and may lead to patient benefit, especially with regard to diarrhoea. This is supported by a recent community study reporting reduced grade 3 or 4 mucocitis and diarrhoea with PK dosing compared to historical controls [
53].
So as to test the generalisability of overall survival and progression free survival reported in the studies of PKA dosing we examined the consistency of reported findings and compared the BSA arms of the comparative PKA versus BSA studies with BSA arms published in the literature (Additional files
6 and
7). It is clear that the BSA arms from the three PKA versus BSA comparative studies were well aligned with BSA arms from relevant studies [
45‐
52] in both overall survival and progression free survival. Available time to event data for PK arms were consistent within their particular dose regimen (FU + FA or FOLFOX6) [
23,
28,
38‐
40]. Estimates of median and mean survival times from well-fitting Weibull models of these studies indicate gains from PK monitoring (Additional file
7).
Unfortunately much of the evidence suggesting that PKA benefits overall and progression-free survival comes from 5-fluorouracil regimens that are now outmoded. These clinical studies have employed either HPLC or the My5-FU immunoassay procedure to estimate plasma 5-fluorouracil. Several studies suggest that a high correlation exists between My5-FU, HPLC and LC-MS/MS methods. Relative to reference assays with HPLC/tandem mass spectrometry the My5-FU immunoassay produced outlying estimates only at low plasma concentrations and with a degree of inaccuracy unlikely to lead to dangerous increases in dose when used in conjunction with suggested algorithms [
27,
43]. On the available evidence, it seems unlikely that when used in conjunction with published dose adjustment algorithms these assays would result in dangerous overdosing.
The place of My5-FU in clinical practice
Our review summarises the available evidence on PK dosing of 5-flourouracil in advanced colorectal cancer patients treated with 5-flourouracil and shows a link to favourable survival outcomes. My5-FU would seem to have a place in dose guidance to either reduce the 5-flourouracil dose and minimise toxicity or to increase the dose to prolong survival by generating high intra-tumour levels of 5-flourouracil that are effective in cancer treatment and prevent loss of response since a clear relationship between 5-flourouracil levels and response could be shown [
27].
Currently, the 5-flourouracil dose given to individual patients to provide a certain response rate and overall survival is defined by different non-PK trials that do not allow dose increase. Therefore, within conventional efficacy, dose increases do not happen in clinical practice. However, if guided by My5-FU, increasing the dose would be a possibility for patients who have acceptable levels of side-effects.
It is important to keep in mind though that non-response to 5-flourouracil is not only related to dosage limitation. The main cause would be inherent resistance to 5-flourouracil rather than dose. Therefore dose adjustment is not going to prevent loss of response in all patients.
It is important that dose increases would be ruled by algorithms as well as clinical judgment. In 2011, Saam reported a US experience with My5-FU suggesting that physicians in practice made larger reductions than increases in 5-flourouracil doses [
54]. And while Gamelin et al. [
38] used an algorithm that allowed 50–70 % dose increases for some patients to reach the 5-flourouracil target range; it appears that physicians not bound to an adaptation protocol generally increased doses by only 10–20 %, illustrating a cautious attitude towards upward dose adjustment [
54]. This might result in PK dose adjustment being less effective in clinical practice than in the research environment because different clinicians may apply dose increases more cautiously than in reported studies.
Under the current knowledge base it is hard to gauge where the My5-FU assay will fit into clinical practice. Successful pharmacokinetic dose adjustment using My5-FU in clinical practice will depend on a) accurate estimation of plasma 5-flourouracil, b) an appropriate algorithm for dose adaptation and c) an appropriate target plasma 5-flourouracil level. No currently available RCT or comparative study used the My5-FU assay for dose adjustment of 5-flourouracil containing chemotherapy regimens. As a result the current knowledge on the accurate estimation of plasma 5-flourouracil relies on comparisons with HPLC. The evidence on algorithms also comes from an indirect comparison with HPLC studies. Furthermore, the only algorithms currently available which have been validated in colorectal cancer patients are based on regimens no longer in clinical practice in the UK [
36,
38] or are unavailable in the public domain [
39]. It is unclear whether the survival gains can be generalised to other treatment regimens that may require alternative and as yet ill-defined adjustment algorithms [
55]. Similarly it is unclear what the optimal plasma 5-flourouracil target level should be. Kaldate et al. [
43] argue that newer extended infusion time regimens which are generally less toxic should use a wider target range of plasma 5-flourouracil levels with the upper limit increased to 30 mg*h/L than the initial target range of 20–24 mg*h/L established for the 8 h 5-flourouracil + folinic acid regimen [
27]. However, no study was identified that made use of this algorithm. Most patients with CRC are nowadays treated with combination therapy the doses of which are defined by clinical trials. If a single agent fluoropyrimidine is needed, then capecitabine is often given. The most common 5-flourouracil combinations are with either irinotecan or oxaliplatin (FOLFIRI/FOLFOX). Even here, there are different types of FOLFOX. This complexity of treatment and modern treatment regimens are not reflected in the available trials on 5-flourouracil PK dosing.
The next step is therefore to evaluate the assay in combination therapy using standard of care regimens in clinical trials. When designing such trial several factors need to be considered: tumour type and combination of drugs, 5-flourouracil scheduling including oral fluoropyrimidines and the genetic makeup of colorectal cancer. Colorectal cancer has been divided into different molecular subtypes based on gene expression profiling. The success of PK dose adjustment is likely to vary by molecular subtype as an association between genotypes and outcomes could be shown [
56‐
58]. With increasing use of genetic profiling, certain cohorts may be defined who would benefit more from PK testing. Understanding molecular biomarkers for predicting 5-flourouracil response could therefore aid appropriate use of PK dose adjustment.
In summary, it is hard to know where the My5-FU assay fits into clinical practice without the results of PK trials using the best available treatment for a given tumour type using conventional or PK dosing.
Strengths and limitations
The main strengths of our study include the rigorous and comprehensive systematic review methodology applied, the comprehensive approach to the available evidence reaching from a highly sensitive search strategy to inclusion of comparative as well as single arm studies of clinical effectiveness of BSA and PKA dosing, the quantitative assessment and modelling of survival outcomes in individual studies, and our effort to assess the consistency of results reported in PKA studies from the perspective of the broader literature. There are several limitations in our review, these stem partly from the fact that the evidence on PKA versus BSA dosing in treating colorectal cancer is weak in both quantity and quality, from the fact that much of the evidence derives from outmoded treatment regimens, from the necessity of reconstructing individual patient data, and from the possibility that included studies may suffer from selective outcome reporting.
Cost-effectiveness of PK dosing
Goldstein et al. [
59] constructed a simple multistate Markov model to assess the cost effectiveness of PKA versus BSA dosing in a FOLFOX regimen. Survival estimates for the PKA arm were based on Capitain et al. [
39] and for the BSA arm on Tournigand et al. [
49]. Incidence of adverse events was taken from Capitain et al. [
39] for the PKA arm and from Höchster et al. [
51] for the BSA arm, and cost estimates were based on US practice. The authors estimated that PKA delivered an extra 1.46 quality adjusted life years at an extra cost $ 37,173. The incremental cost effectiveness ratio of $22,695/QALY was robust in univariate and multivariate sensitivity analyses. The authors concluded that at a $50,000/QALY threshold PK FOLFOX is cost effective for metastatic colorectal cancer and that it should be further evaluated in comparative effectiveness studies.