High-dose infusions of MTX, a folate antagonist blocking intracellular dihydrofolate reductase and thymidylate synthetase, are used widely in the treatment of malignancies but carry a risk of significant toxicity, the causes of which need to be clarified. In this retrospective study of the role of serum albumin in 325 HDMTX 5g/m2 courses given to children with ALL we found that hypoalbuminemia was a frequent occurrence, apparently caused by preceding asparaginase therapy. There was a clear relation to recent doses of ASP and to the intensity of ASP therapy, hypoalbuminemia being less frequent in SR/IR children receiving reduced number of doses and in HR children with longer intervals between doses. A low serum albumin, especially when < 30 g/L, was associated with lower MTX clearance and had a marked impact on end-of-infusion MTX concentrations that were very high in one third of infusions. Creatinine rises did not occur more frequently, and MTX elimination was not delayed. Later systemic toxicity, however, was more pronounced with a doubled risk of stomatitis.

This study is retrospective and therefore subject to several limitations. Clinical information in records may have been inaccurate and incomplete, and some laboratory values were missing. Accurate grading of stomatitis was not possible retrospectively. Information on concurrent infection, concomitant medication and management of fluid balance was not collected. Deviations from protocol guidelines were not registered, so that strict adherence cannot be assumed. Presence of known risk factors for DME could not be recorded systematically. Therefore, the interaction between hypoalbuminemia and other factors could not be elucidated, and multifactorial modelling was not possible. Thus, the study is a single-factor analysis. It is a strength in interpreting the results, however, that the study cohort is composed of children with the same diagnosis receiving HDMTX at the same dose according to a specified guideline. Also, the analytic design with division into four groups permitted exploration of the importance of the degree of hypoalbuminemia. Overall, we think that our data material is sufficiently robust for a descriptive analysis that can provide reliable findings.

Low serum albumin is encountered quite frequently in children treated for cancer. At the time of diagnosis hypoalbuminemia may be present in half the children, and if severe may be associated with reduced survival [28]. Later during chemotherapy, a low serum albumin often can be attributed to well-known complications: malnutrition or hepatotoxicity quite commonly, renal or intestinal loss occasionally. In previous studies in adults, hypoalbuminemia has been reported in connection with 18% to 27% of HDMTX courses [15,16,17,18,19]. In two childhood series from low-middle-income countries 32% and 35% of the children were hypoalbuminemic, possibly because of malnutrition [21, 22].

In the present study the frequency of reduced serum albumin was close to 50%. The cause of this high prevalence is not immediately obvious. We found, however, a clear relation to recent administration of ASP. A single dose of L-asparaginase causes depletion of L-asparagine for two weeks, blocking lymphoblast replication, but also reducing synthesis of various proteins including albumin [24, 25]. ASP therapy in the NOPHO ALL-2008 protocol was intensive and prolonged, most children receiving 15 doses at 2-week intervals after induction. During consolidation hypoalbuminemia was present in up to 85% of HDMTX courses, often with values < 30 g/L. During maintenance, hypoalbuminemia was still frequent in children receiving ASP at 2-week intervals, less frequent and severe in those receiving ASP at 6-week intervals, and absent in those with inactivation of ASP. After cessation of ASP therapy, albumin levels were normal. Thus, it appears that ASP is responsible for the hypoalbuminemia, which is a biomarker of ASP activity, previously reported to influence the clearance of dexamethasone [25].

Hypoalbuminemia may influence the pharmacokinetics of MTX in several ways [19, 29]. MTX is partially bound to albumin in the intravascular compartment. Approximately 70% is cleared by the kidneys, mostly by glomerular filtration, to a smaller extent by tubular secretion. Free MTX is distributed in the extravascular space and taken up by cells. Hypoalbuminemia reduces the amount of bound MTX and increases the free fraction. Decreased osmotic pressure reduces blood volume and causes renal hypoperfusion with reduced glomerular filtration of MTX. Our data showed that pre-infusion serum creatinine values rose with falling serum albumin, supporting the presence of reduced filtration [30]. Low serum albumin increases the fraction of unbound MTX, resulting in greater penetration into “deep” less vascular tissue spaces (V3) that are cleared more slowly. A very low albumin may cause formation of third spaces (pleural effusion, ascites) which may retain MTX for prolonged periods. No such cases were seen in this series.

Renal hypoperfusion seems a plausible explanation for the most striking finding in this study: the impact of the serum albumin level on the steady-state MTX concentration. The median end-of-infusion MTX concentration rose markedly from less than 75 µM in group D to more than 125 µM in groups B and A, and the risk of a very high MTX-h23 ≥ 150 µM tripled from around 10% to more than 30%. This is consistent with reduced MTX elimination depending on the level of serum albumin and corresponds well with the MTX clearances that were determined, falling from an average of close to 11 L/h/1.73 m2 in group D to 9 L/h/1.73 m2 in group A.

A high steady-state MTX concentration is predictive of DME, albeit less predictive than a 50% rise in serum creatinine [8]. We therefore expected to see increased occurrence of creatinine rises and slow elimination in the hypoalbuminemic groups – but none was seen. At 48 h after starting the infusions the differences in serum MTX concentrations had disappeared. The explanation for this accelerated elimination in the hypoalbuminemic groups is not obvious, but prompt institution of hyperhydration 4500 ml/ m2/day with increased alkalinization may have facilitated faster elimination, and faster filtration of the high free fraction may have contributed. It can also be noted that eGFR tended to rise after completing the infusion, especially in groups A and B.

Significant creatinine rises occurred in 10.5% of the courses, oddly with the lowest rates in courses with very low serum albumin. The overall frequency corresponds well with previously reported incidences of 9–18.5% [8, 15, 17, 18, 21, 31]. DME has been defined in various ways in different reports, but it seems clinically reasonable to consider an elimination time > 72 h as prolonged; this occurred in close to 10% of the courses, compared with 17% of 4970 HDMTX treatments in the NOPHO ALL 2000 protocol [32].

While hypoalbuminemia did not delay MTX elimination, the systemic toxicity appearing after the courses was somewhat increased. Oral stomatitis occurred almost twice as frequently in the groups A and B than in groups C and D; a similar difference has been reported in a recent study [22]. The impact on myelotoxicity was less pronounced, with modest increases in low nadir values. The greater toxicity probably stems from increased intracellular accumulation of MTX and formation of polyglutamates during the infusion because of higher free MTX concentrations. It is also in agreement with the higher AUC values, indicating increased MTX exposure.

The clinical consequences of the findings remain to be determined. Strictly, apart from stomatitis the increased MTX concentrations during the infusions had a rather limited effect on the toxicity. It seems prudent, however, to attempt to reduce the risk of reaching a very high MTX concentration during the infusion when serum albumin < 30 g/L. MTX dose reduction would seem rational but may not be necessary: the reduction in clearance and the rise in AUC after all are not very pronounced, GFR appears to become normal during hyperhydration, and MTX elimination is not delayed. An infusion of albumin before starting MTX also seems tempting but may not be recommended because of cost and side effects [16] and probably should be reserved for children with edema, pleural effusion, or ascites. Extension of pre-hydration from 4 to 12 h could be tried; it gave no benefit in a randomized trial [31], but it has been standard practice in Finland and resulted in rapid clearance after 42 or 48 h in 52% of courses [33], compared to 29% in the present series. Hyperhydration with 4500 ml/m2/day from the start of the infusion or increased amount of bicarbonate in the hydration fluid also reasonably could be tried. Measuring serum MTX during the infusion to adjust the infusion rate is another option, but it may not decrease the risk of toxicity [34, 35]. In addition, aiming to avoid stomatitis, an extra folinic acid dose after 36 h or a double dose at 42 h could be considered, and folinic acid mouth rinses could be started early. Under all circumstances, extra care is warranted when starting HDMTX courses with a low serum albumin: fluid balance and urine pH should be monitored meticulously, and nephrotoxic drugs or medications competing for tubular excretion strictly avoided [10].

The impact of hypoalbuminemia on the course of MTX infusions may be overcome, but the relation to ASP therapy raises a question which may have wider repercussions. In current international ALL protocols many HDMTX courses are given while ASP therapy is ongoing. Hypoalbuminemia can be seen as an indicator that ASP has caused effective L-asparagine depletion. The enzyme catalyzes the hydrolysis of L-asparagine to aspartic acid and ammonia, depleting serum and CSF of L-asparagine. Unlike normal cells, lymphoblasts are deficient in asparagine synthetase, and they are therefore “starved” by the depletion, entering resting phase G0-G1 with inhibition of S-phase DNA synthesis [36]. With lymphoblasts forced into metabolic arrest, the effect of antimetabolic drugs like MTX may be antagonized [37]. Thus, it may be speculated that ASP impairs the antileukemic effect of HDMTX while still causing systemic toxicity. It has previously been reported that concomitant administration of ASP has little effect on the toxicity of HDMTX but reduces formation of MTX polyglutamates in cells [38]. Clarification of these issues is of considerable importance.

In conclusion, this descriptive review of HDMTX courses in the treatment of children with ALL has shown that hypoalbuminemia is a frequent occurrence when therapy with ASP is ongoing, and that it has a significant impact on MTX pharmacokinetics, depending on the degree of albumin suppression. We suggest that a level < 30 g/L should be considered clinically significant, defining a group with low initial MTX clearance, high steady-state concentration, high MTX exposure, and doubled risk of stomatitis. Hyperhydration may increase MTX elimination when the end-of-infusion concentration is high, but toxic tissue accumulation may already have occurred, and the utility of an albumin infusion prior to the MTX infusion seems worth investigating. Furthermore, disadvantages from scheduling HDMTX in close proximity to ASP doses merit further deliberation.