We have read with great interest the letter “Increased lipoprotein(a) levels are not a steady prothrombotic defect” by Korte et al, recently published in Blood.1 The authors report on a small case series of 7 pediatric patients treated with asparaginase either because of acute lymphoblastic leukemia (acute lymphoblastic leukemia [ALL]-Berlin-Frankfurt-Münster [BFM] 95 trial) or because of non-Hodgkin lymphoma (non-Hodgkin lymphoma [NHL]-BFM trial). In the longitudinally followed-up patients, mean lipoprotein (Lp)(a) concentrations decreased by approximately 75% during the course of polychemotherapy, comparing values on day 8 with concentrations on day 22 (P = .049; paired t test). The results obtained were discussed as mainly asparaginase-induced. Therefore, Korte et al conclude that elevated Lp(a) concentrations could not be assumed to be a stable risk marker of venous thromboembolism (VTE) during chemotherapy. Moreover, they suggest that increased Lp(a) levels should not be looked upon as a “prothombotic defect” since a clearly defined pathophysiological model on how increased Lp(a) concentrations can convey an increased prothrombotic risk is still missing.1
The first of our studies not correctly cited by Korte et al enrolled patients with venous thrombosis. The rate of recurrence with respect to isolated or combined prothrombotic risk factors was investigated. Lp(a) was included as a potential risk factor of recurrent venous thrombosis because high levels of Lp(a) have already been shown to be a risk factor for first manifestation of cardiovascular disease4 and venous thromboembolism in white children and adults.5-7 In this study, no patient was suffering from acute leukemia or being treated with asparaginase. Moreover, we could definitely not identify an elevated level of Lp(a) as an isolated risk factor of recurrent venous thrombosis: the relative risk of suffering a recurrent venous thrombosis in patients with elevated levels of Lp(a) was 2.6 (95% confidence interval 0.5-33.3) (both incorrectly stated by Korte et al).2
In the second study mentioned by the authors, we prospectively performed thrombophilia screening in children with acute lymphoblastic leukemia treated according to the BFM 90/95 trials (but not patients with non-Hodgkin lymphoma, as was incorrectly stated by Korte et al) with respect to symptomatic venous thromboembolism as the study end point.3 We found the rate of prothrombotic risk factors in the total patient group to be within the range of the healthy population. However, 46% of the children with at least one thrombophilic genetic risk (n = 58; factor V G1691A and factor II G20210A mutation; protein C, protein S, or antithrombin deficiency; homozygous MTHFR T677T genotype; high level of Lp(a)) suffered symptomatic venous thrombosis during the study period. The children affected with symptomatic vascular accidents and suffering from elevated Lp(a) either alone (n = 2) or in combination with further of the mentioned risk factors (n = 6) received Escherichia coli asparaginase. Therefore, in the “Discussion” section we clearly pointed out that, besides the genetic risk factors of thrombophilia, additional factors such as endothelial cell injury or further acquired coagulation imbalance commonly described during combined steroid asparaginase administration may function as trigger mechanisms for early thrombotic manifestation during childhood ALL (BFM-adapted protocols).3 This has been ignored or overlooked by Korte et al.1
In addition to the aspects mentioned above, we would like to comment on some other points raised in the letter by Korte et al1: Lp(a), first identified by Berg in 1993,8is strongly genetically determined, and its apo(a) phenotypes account for more than 90% of the variation in Lp(a) plasma concentrations.9-11 Different properties of antibodies and calibrators used to determine Lp(a) have led to a wide range of values that are not comparable between different methods and laboratories.12 Therefore, the cutoff value indicating an increased prothrombotic risk is not transferable between different populations and studies. The letter published by this group provides no information on which Lp(a) method is used in their setting. Therefore, comparison between our values and the concentrations of the 7 cases is meaningless.12 Their findings are, however, in accordance with previously published data on the influence of asparaginase during the course of leukemia induction therapy in BFM-adapted treatment protocols. The decrease is similar to that shown for plasminogen and further coagulation proteins containing asparagine or glutamine as amino acid, for example, the targets of asparaginase.13 The percentage decrease is, however, clearly dose- and source-dependent.13-15 Unfortunately, Korte et al provide no information on the source and dosage of asparaginase administered to the children, the asparaginase activity, or the percentage of asparagine depletion concomitant with Lp(a) value.
We also would like to refer to the current underlying pathophysiological models on how elevated Lp(a) functions as a prothrombotic risk factor. There is a striking similarity between theapo(a) gene and plasminogen, which share a partially identical amino acid sequence.16 It has been demonstrated by Harpel et al17 that Lp(a) inhibits the binding of plasminogen to plasmin-modified immobilized fibrinogen, thus indicating that both molecules compete for similar lysine-binding sites, providing a potential mechanism to explain the association between thrombosis, coronary atherosclerosis, and elevated blood concentrations of Lp(a).17
From the statistical point of view, we have to add that Lp(a) concentrations usually follow a clearly biased but not-normal distribution. Thus, the appropriate statistical method to compare Lp(a) levels of patients followed up longitudinally is the Wilcoxon signed rank test/Friedman test instead of the pairedt test, which is to be used for normal distributed data only.18
In conclusion, as shown in several studies, an elevated Lp(a) concentration is an independent risk factor for cardiovascular disease as well as venous thromboembolism in patients suffering from spontaneous thrombosis or from vascular accidents additionally triggered by underlying diseases. The 7 cases presented by Korte et al in the September 15 issue of Blood raise methodological as well as statistical problems and should therefore not be compared with previously published articles. Finally, it is not justified from 7 cases to draw any conclusion on whether or not elevated Lp(a) is a prothrombotic risk factor.
Elevated lipoprotein(a) concentration is an independent risk factor of venous thromboembolism
We thank Nowak-Göttl et al for their response, and we appreciate their discussion of this issue. The aim of our report1-1 was to bring attention to the fact that lipoprotein (Lp)(a) levels can be modulated through disease processes as well as pharmacological interventions. If one accepts that an increase in Lp(a) conveys a risk for thromboembolism, it seems logical that diseases or interventions that decrease Lp(a) should reduce this (ie, the Lp(a)-associated) risk. Given the numerous issues that Nowak-Göttl et al bring up with their rebuttal letter, we would like to focus on those that seem important to us.
First, the statement with regard to their 2001 paper1-2 is very confusing to us. Nowak-Göttl et al state that they could not identify an elevated Lp(a) level as an “isolated” risk factor, whereas the title of the rebuttal letter suggests that increased Lp(a) concentrations are an “independent” risk factor. Besides, Nowak-Göttl et al incorrectly cite their paper with regard to the 95% CI for the Lp(a)-associated relative risk.
Second, given their results with acute lymphoblastic leukemia (ALL) patients1-3 and considering our own experience,1-4 we agree with Nowak-Göttl et al's rebuttal comment that genetic factors cannot be the only risk factors for venous thromboembolism (VTE) during ALL therapy. However, this comment is contradictory to the title of the rebuttal letter. An independent risk factor, by definition, is sufficiently strong on its own to induce the change under investigation. Also, given Nowak-Göttl et al's above comment, subordination of increased Lp(a) levels under the subtitle “Prevalence of single established prothrombotic risk factors” seems incorrect since it hasnot been proven to be a single established risk factor in this setting (ALL therapy). Finally, and interestingly enough, the Kaplan-Meier plot in the 1999 paper3 (Fig 1)shows that most thromboses occurred between treatment days 20 and 30—exactly the time frame in which we found the most pronounced decrease in Lp(a) during asparaginase therapy, suggesting that the potential thrombogenic influence of Lp(a) wasdecreased at this point.
Third, Nowak-Göttl et al criticize our sample size, but the sample was large enough to prove significant intra-individual changes. Also, the criticism seems doubtful, given that Nowak-Göttl et al base their discussion regarding Lp(a) and thrombosis in ALL therapy on a total of 8 patients1-3; moreover, only 2 of those had no other prothrombotic risk factors identified.
Fourth, our samples were measured with a commonly used nephelometric assay (Dade Behring, Hamburg, Germany), and we agree that standardization is an important issue.1-5 However, methodology is a different thing and not an issue with our results, since all intra-individual changes were measured with the same method.
Fifth, Lp(a) (described and investigated by Berg much earlier than 1993, namely, in the 1960s and 1970s) resembles plasminogen and competes with plasminogen for binding sites in vitro. If Nowak-Göttl et al accept this model, it is not conceivable why they would not accept that this competition will diminish if Lp(a) levels decrease.
Sixth, Lp(a) often shows a skewed distribution, but this obviously depends on the population investigated. Our sample was normally distributed, as proven by Kolgomorov-Smirnov testing. However, a significant difference was also obtained using the (nonparametric) Wilcoxon signed rank test.
In conclusion, we cannot concur with Nowak-Göttl et al that our results were biased by methodological or statistical problems. Given the current knowledge, we can also not concur that an elevated Lp(a) concentration is a stable risk factor for VTE at any time; data to support this are yet to be generated. Although our patient sample was small, it was large enough to demonstrate that Lp(a) concentrations can be modulated during asparaginase-containing therapy, as also shown by others. Disease processes and pharmacological interventions can modify Lp(a) levels and, thus, the risk for VTE related to Lp(a) levels. This needs to be taken into account when risk stratification for VTE on the basis of Lp(a) concentrations is performed.