Fluorouracil chemotherapy: The My5‑FU assay for guiding dose adjustment

6 Considerations

6.1 The Diagnostics Advisory Committee reviewed the evidence available on the clinical and cost effectiveness of the My5‑FU assay for guiding dose adjustment in patients having 5‑fluorouracil (5‑FU) chemotherapy by continuous infusion. The Committee noted that the evidence for clinical effectiveness included: studies and manufacturer validation data that compared the My5‑FU assay with the analytical reference standard technologies (high‑performance liquid chromatography and liquid chromatography‑mass spectrometry); studies that reported algorithms developed to facilitate pharmacokinetic dose adjustment of continuous infusion 5‑FU; and studies that reported clinical outcomes in patients with colorectal cancer who received either pharmacokinetic dose adjustment or body surface area dosing.

6.2 The Committee discussed whether the My5‑FU assay could be considered equivalent to high‑performance liquid chromatography and liquid chromatograph‑mass spectrometry for the quantitative determination of 5‑FU in plasma. The Committee noted that the available comparative data appeared to show that despite high correlation between the My5‑FU assay and high‑performance liquid chromatography or liquid chromatography‑mass spectrometry, there was substantial variability between the methods, particularly with regard to the imprecision of the assay in its lower measuring range. The Committee considered that in clinical practice, patients who had under‑dosing would be likely to have 5‑FU plasma levels that would fall within the lower end of the My5‑FU assay's measuring range. This imprecision could therefore impact on the My5‑FU assay's reliability for clinical decision‑making, specifically it may be less likely to detect patients who have low 5‑FU plasma levels and who would potentially benefit from an increased dose of 5‑FU at the next cycle. The Committee therefore concluded that it was not appropriate to consider the My5‑FU assay equivalent to high‑performance liquid chromatography and liquid chromatography-mass spectrometry for determining plasma levels of 5‑FU and guiding dose adjustment in clinical practice. The Committee noted that this conclusion introduced substantial uncertainty into the interpretation of both the clinical outcome data (based mainly on studies using high‑performance liquid chromatography) and the cost‑effectiveness modelling.

6.3 The Committee discussed the published dose adjustment algorithms that had been included in the clinical‑effectiveness review, and noted that 3 dose adjustment algorithms had been developed for use in patients having 5‑FU chemotherapy for colorectal cancer and 1 for patients having 5‑FU chemotherapy for head and neck cancer. The Committee heard from clinical specialists that only 1 of the published dose adjustment algorithms, which is based on a 5‑FU + folinic acid + oxaliplatin (FOLFOX) regimen for colorectal cancer (Kaldate et al. 2012), could be considered applicable to current practice. The Committee considered whether the target range that had been established by Kaldate et al. (2012) could be extrapolated to head and neck, stomach, and pancreatic cancer. The Committee noted that, although it was plausible that the dose‑outcome relationship suggested for people with colorectal cancer may be applicable to people with other types of cancer, no data were available to support this assumption. The Committee concluded that it was uncertain whether target ranges, and their associated dose adjustment algorithms, were transferrable between cancer types.

6.4 The Committee considered both the applicability and quality of the studies included in the colorectal cancer clinical outcomes analysis. The Committee noted that Gamelin et al. (2008) is a randomised controlled trial and considered that, despite using an outdated chemotherapy regimen, it may provide more robust survival estimates than Capitain et al. (2012), a retrospective study that reported results from a FOLFOX regimen. However, the Committee also noted that the External Assessment Group had identified concerns with the study design reported by Gamelin et al. (2008), in particular, the methods of randomisation were uncertain and it was not clear whether patients and investigators were blinded to allocation. The Committee considered that the studies included in the analysis could be regarded as 'proof of concept' studies, which demonstrated that the use of pharmacokinetic dose adjustment in the treatment of colorectal cancer was both feasible and had the potential to improve outcomes. However, the Committee concluded that these studies did not provide sufficiently robust effect estimates to determine whether pharmacokinetic dose adjustment was clinically effective compared with body surface area dosing, and noted that it was uncertain whether adjusting doses of 5‑FU would translate into an improvement in clinical outcomes.

6.5 The Committee discussed the overall and progression‑free survival data that had been included in the colorectal cancer clinical outcomes analysis and noted that the limited outcome data available for this comparison were largely drawn from Kaplan–Meier curves that had been reconstructed and modelled by the External Assessment Group. The Committee considered that median survival estimates from each of the included studies indicated a trend towards increased progression‑free and overall survival in patients who received pharmacokinetic dose adjustment, but noted that the clinical and statistical significance of the reported increases were uncertain and that the effect estimates obtained from the modelled Kaplan–Meier curves were open to substantial bias because of incomplete reporting of survival data in the included studies. Additionally, the Committee noted that 2 studies (Kline et al. 2013 and Patel et al. 2014) used the My5‑FU assay to measure 5‑FU plasma levels, although Patel et al. did not report survival data. The Committee also heard from clinical specialists that the overall survival estimates reported for pharmacokinetic dose adjustment (19 to 28 months) did not appear to be representative of current clinical practice. The Committee concluded that, because of potential biases from both study designs and the use of reconstructed survival data, there was substantial uncertainty around the magnitude of the effect of pharmacokinetic dose adjustment on progression‑free and overall survival.

6.6 The Committee considered the toxicity data included in the colorectal cancer clinical outcomes analysis and discussed the likely impact of pharmacokinetic dose adjustment on toxicities associated with 5‑FU chemotherapy. The Committee questioned which toxicities reported in the analysis were likely to be dose‑dependent and heard from a clinical specialist that cardiac toxicities were unlikely to be related to 5‑FU dosing. The Committee considered that the available data suggested pharmacokinetic dose adjustment may result in a significant reduction in the number of patients who experience diarrhoea, but that the impact of pharmacokinetic dose adjustment on other side effects was uncertain. Additionally, the Committee noted that the lack of blinding in the included studies, combined with the subjective nature of side effect reporting, could have introduced bias into the reported effect estimates. The Committee heard from patient experts that side effects associated with continuous infusion 5‑FU may have a significant impact on a patient's quality of life and that an intervention that reduced the severity or duration of these side effects could have a substantial impact. The Committee also heard from clinical specialists that some toxicities have a greater impact than others and that most of the hospital admissions associated with 5‑FU toxicity are as a result of diarrhoea or neutropenic sepsis. The Committee concluded that although pharmacokinetic dose adjustment appeared to reduce the incidence of diarrhoea, there was insufficient evidence to determine whether it would have a substantial impact on other toxicities that may be associated with a negative impact on quality of life.

6.7 The Committee noted that no clinical outcome data were found for pharmacokinetic dose adjustment in people having continuous infusion 5‑FU for pancreatic or stomach cancer. The Committee concluded that there was insufficient evidence to make recommendations on the use of the My5‑FU assay in these populations.

6.8 The Committee considered that, because of a lack of data, the External Assessment Group had not been able to do any subgroup analyses. The Committee concluded that it was not possible to determine whether differential effects associated with pharmacokinetic dose adjustment would be observed in people with DPD (dihydropyrimidine dehydrogenase) deficiency, people with impaired renal or liver function, people whose body surface area is outside the standard range for dosing 5‑FU and people with a less favourable performance status.

6.9 The Committee considered the cost‑effectiveness analysis and noted that the economic model included 2 base‑case analyses. The first was based on progression‑free and overall survival data from a randomised controlled trial (Gamelin et al. 2008), which used a 5‑FU + folinic acid regimen. The second was based on progression‑free and overall survival data from a retrospective study (Capitain et al. 2012), which used a FOLFOX6 regimen. The Committee discussed the incremental cost‑effectiveness ratios (ICERs) for both deterministic base cases and noted that the My5‑FU assay appeared to be cost effective (£4148 per quality‑adjusted life year [QALY] gained in the FOLFOX6 base‑case analysis and £5853 per QALY gained in the 5‑FU + folinic acid base‑case analysis). However, the Committee noted that the results of the deterministic base‑case analyses were based on the assumption that the effectiveness of the My5‑FU assay is similar to high‑performance liquid chromatography, and that the effect estimates obtained from Gamelin et al. (2008) and Capitain et al. (2012) would be observed in routine clinical practice. The Committee considered that, because of the bias observed in the lower measuring range of the assay and the uncertainty associated with the relative survival estimates for pharmacokinetic dose adjustment, these assumptions would be unlikely to be realised in clinical practice. On this basis, the Committee concluded that the FOLFOX6 and 5‑FU + folinic acid deterministic base‑case ICERs were subject to substantial uncertainty.

6.10 The Committee considered the results of the base‑case sensitivity analyses and noted that the cost effectiveness of the My5‑FU assay was dependent on increased overall survival being realised in practice, because the reduction in toxicities alone was not sufficient to offset the increased costs associated with the My5‑FU assay in the economic model. When the relative progression‑free and overall survival effect estimates were removed from the economic model, the resulting ICERs were £435,819 per QALY gained in the FOLFOX6 analysis and £435,804 per QALY gained in the 5‑FU + folinic acid analysis.

6.11 The Committee considered the threshold analyses for both the FOLFOX6 and 5‑FU + folinic acid base cases and noted that the analyses showed that the use of the My5‑FU assay could be considered cost effective if the small overall survival gains (hazard ratios of between 0.85 and 0.98) could be achieved in clinical practice. The Committee considered the uncertainty in the relative survival estimates obtained from Gamelin et al. (2008) and Capitain et al. (2012) and noted that the My5‑FU assay could not be considered similar in effectiveness to high‑performance liquid chromatography and liquid chromatography‑mass spectrometry because of uncertainty in the precision of the My5‑FU assay in its lower measuring range. The Committee also considered that the uncertainty in the clinical‑effectiveness data, because of bias in the design of the Gamelin et al. (2008) and Capitain et al. (2012) studies, could not be fully captured in either the probabilistic sensitivity analyses or univariate sensitivity analyses, and therefore the resulting ICERs were likely to be subject to substantial uncertainty. The Committee therefore concluded that the uncertainty associated with the reported ICERs was too great to conclude that the use of the My5‑FU assay would be cost effective in routine clinical practice.

6.12 The Committee considered the External Assessment Group's indicative economic analysis for head and neck cancer. The Committee noted that, although the relative survival gains needed for the My5‑FU assay to be considered cost effective were relatively small (hazard ratio for overall survival: 0.990), the uncertainties associated with the clinical effectiveness of the My5‑FU assay in people with head and neck cancer meant that the results of the analyses were highly uncertain.

6.13 The Committee considered the likely impact of a reduction in toxicities associated with continuous infusion 5‑FU in practice. The Committee heard from patient experts that side effects are often cumulative and increase during a course of chemotherapy, and clinical specialists suggested that it is not always possible to predict which patients will experience toxicities after the first cycle of chemotherapy. The Committee heard from clinical specialists that in current practice, patients who experience severe toxicities may have their dose of 5‑FU reduced but that most of the side effects can often be treated with additional medications. The Committee considered that the most severe toxicities are often experienced by patients who have DPD deficiency and heard from clinical specialists that approximately 1 in 300 patients having continuous infusion 5‑FU are thought to die within 30 days of their first chemotherapy cycle as a result of severe 5‑FU toxicities. The Committee concluded that the incidence of severe side effects that become apparent during the first cycle of continuous infusion 5‑FU was unlikely to be reduced by pharmacokinetic dose adjustment, but that an impact on less severe toxicities was more likely to be achieved in practice.

6.14 The Committee considered the likely impact of pharmacokinetic dose adjustment on the doses of 5‑FU that would be administered in practice. The Committee noted that the studies included in the clinical‑effectiveness review tended to show that most patients needed dose increases as a result of measuring 5‑FU plasma levels, and heard from clinical specialists that pharmacokinetic dose adjustment could result in a greater number of patients having an optimal therapeutic dose of 5‑FU because current practice does not identify patients who metabolise 5‑FU at an increased rate and who consequently receive doses that have a reduced therapeutic effect. However, the Committee concluded that, at present, this may not be achieved if the My5‑FU assay were implemented into clinical practice because of concerns regarding the imprecision of the assay within the clinical measuring range and insufficient evidence to demonstrate to clinicians that increasing the dose for patients who do not experience toxicities could result in improved clinical outcomes.

6.15 The Committee heard from patient experts that they would welcome the opportunity to receive tailored dosing of continuous infusion 5‑FU and believed that any inconvenience caused by 5‑FU plasma monitoring, such as increased outpatient attendances, would be offset if quantity and quality of life were improved. The Committee considered that the most notable benefit associated with pharmacokinetic dose adjustment of continuous infusion 5‑FU was its potential to increase the number of people having optimal therapeutic doses without increasing toxicities, but concluded that further research was needed to confirm whether this would be achieved in practice.

6.16 The Committee acknowledged that many clinicians now prescribe capecitabine as an alternative to 5‑FU and noted that the My5‑FU assay is not licensed for use with capecitabine. The Committee heard from clinical specialists that around 30‑40% of colorectal cancer patients currently receive continuous infusion 5‑FU and that recently licensed biological agents are marketed for use in conjunction with continuous infusion 5‑FU. The Committee concluded that it was likely that there will continue to be a significant proportion of patients who receive continuous infusion 5‑FU, and who may benefit from pharmacokinetic dose adjustment in the future.