On the prediction of Hodgkin lymphoma treatment response

Purpose. The cure rate in Hodgkin lymphoma is high, but the response along with treatment is still unpredictable and highly variable among patients. Detecting those patients who do not respond to treatment at early stages could bring improvements in their treatment. This research tries to identify the main biological prognostic variables currently gathered at diagnosis and design a simple machine learning methodology to help physicians improve the treatment response assessment. Methods. We carried out a retrospective analysis of the response to treatment of a cohort of 263 Caucasians who were diagnosed with Hodgkin lymphoma in Asturias (Spain). For that purpose, we used a list of 35 clinical and biological variables that are currently measured at diagnosis before any treatment begins. To establish the list of most discriminatory prognostic variables for treatment response, we designed a machine learning approach based on two different feature selection methods (Fisher’s ratio and maximum percentile distance) and backwards recursive feature elimination using a nearest-neighbor classifier (k-NN). The weights of the k-NN classifier were optimized using different terms of the confusion matrix (true- and false-positive rates) to minimize risk in the decisions. Results and conclusions. We found that the optimum strategy to predict treatment response in Hodgkin lymphoma consists in solving two different binary classification problems, discriminating first if the patient is in progressive disease; if not, then discerning among complete and partial remission. Serum ferritin turned to be the most discriminatory variable in predicting treatment response, followed by alanine aminotransferase and alkaline phosphatase. The importance of these prognostic variables suggests a close relationship between inflammation, iron overload, liver damage and the extension of the disease (AU)

No disponible

On the prediction of Hodgkin lymphoma treatment response.

PURPOSE: The cure rate in Hodgkin lymphoma is high, but the response along with treatment is still unpredictable and highly variable among patients. Detecting those patients who do not respond to treatment at early stages could bring improvements in their treatment. This research tries to identify the main biological prognostic variables currently gathered at diagnosis and design a simple machine learning methodology to help physicians improve the treatment response assessment. METHODS: We carried out a retrospective analysis of the response to treatment of a cohort of 263 Caucasians who were diagnosed with Hodgkin lymphoma in Asturias (Spain). For that purpose, we used a list of 35 clinical and biological variables that are currently measured at diagnosis before any treatment begins. To establish the list of most discriminatory prognostic variables for treatment response, we designed a machine learning approach based on two different feature selection methods (Fisher's ratio and maximum percentile distance) and backwards recursive feature elimination using a nearest-neighbor classifier (k-NN). The weights of the k-NN classifier were optimized using different terms of the confusion matrix (true- and false-positive rates) to minimize risk in the decisions. RESULTS AND CONCLUSIONS: We found that the optimum strategy to predict treatment response in Hodgkin lymphoma consists in solving two different binary classification problems, discriminating first if the patient is in progressive disease; if not, then discerning among complete and partial remission. Serum ferritin turned to be the most discriminatory variable in predicting treatment response, followed by alanine aminotransferase and alkaline phosphatase. The importance of these prognostic variables suggests a close relationship between inflammation, iron overload, liver damage and the extension of the disease.

[Spinal cord infarction and recurrent venous thrombosis in association with estrogens and the 20210A allele of the prothrombin gene]. / Mielopatía aguda isquémica y trombosis venosa recurrente asociada a estrógenos y alelo 20210A del gen de la protrombina.

The acute spinal cord infarction is a rare cause of acute-onset paraplegia. Furthermore, it is specially uncommon that the infarction occurs in patients with apparent low predisposition to vascular disease. The 20210A allele of the prothrombin gene (causing a threefold-increased risk in venous thromboembolism) was recently associated with unexplained spinal cord infarction in young women under treatment with estrogens (contraceptive pill). We report a case of anterior spinal artery syndrome resulting from an ischaemic infarction at the anterior aspect of the spinal cord in a healthy 50-year-old woman, carrying this mutation, being the first published case under treatment with transdermal estradiol. She referred the typical sudden-onset back pain associated to clinical anterior spinal artery syndrome with sphincter dysfunction and nontraumatic paraplegia. A possible multiple sclerosis was ruled out and the steroids or immunoglobulin therapy induced no clinical improvement. Cerebrospinal fluid and other investigations were all negative. Sequential MRI scans revealed development of spinal cord infarction from T10 to T11, with increased signal in T2-weighted image (T2). Because she referred a previous thrombophlebitis and suffered a deep-vein thrombosis one month after paraplegia, a complete coagulation study was performed. Antithrombin, proteins C and S, homocysteine, factor V Leiden, lupus anticoagulant and anticardiolipin antibodies were all normal or negatives. In opposite, the 20210A variation was positive (heterozygous) and the factor VIIIc level was very high (280 U/dl eight months later). We argue the relative importance of both findings. The patient had no a substantial recovery over a period of 20 months.Certainly, the prothrombin 20210A seems to be associated with unexplained ischemic myelopathy among the young women with estrogens.

Risk of venous thromboembolism associated with the insertion/deletion polymorphism in the angiotensin-converting enzyme gene.

The circulating levels of angiotensin I-converting enzyme (ACE) are linked with a 287-base pair insertion/deletion (I/D) polymorphism at intron 16 of the ACE gene. Thus, the homozygous deletion (D/D genotype) could cause chronic vasoconstriction, arterial hypertension and, possibly, coronary artery disease. Also, the increase in plasminogen activator inhibitor-1 level and impaired fibrinolysis were related with the D/D genotype. The D allele has been recently associated with venous thrombosis among African-American men as well as among patients that underwent elective total hip replacement. We assess the risk of venous thromboembolism (VTE) linked with each genotype of the I/D ACE gene polymorphism in a Caucasian population by means of a case-control study. We genotyped the ACE gene in a series of 148 patients aged 45.0 +/- 16.0 years (range, 11-80 years), objectively diagnosed in our centre of deep-vein thrombosis or pulmonary embolism, and in 240 thrombosis-free subjects (25-75 years) from the same geographic area. The observed difference in D allele frequencies between patients (0.56) and controls (0.62) was nonsignificant overall; however, statistical significance (P = 0.05) was found by considering the recessive hypothesis (D/D versus I/ D + I/I) [odds ratio (OR) = 0.64, 95% confidence interval (CI95) = 0.42-0.99]. The OR was 0.88 (CI95 = 0.51-1.53; P = 0.65) for the dominant hypothesis (D/D + I/D versus I/I genotypes). The relative risk for the D allele was close to 1 for the dominant hypothesis, both considering gender and recurrent tendency; however, it was protective in men regarding the recessive hypothesis (OR = 0.53, CI95 = 0.29-0.97, P = 0.04). The I/D ACE allele distribution was similar among the 46 thrombophilic patients (antithrombin, protein C or protein S deficiencies, factor V R506Q, factor II G20210A or lupus anticoagulant). In conclusion, among (Spanish) Caucasians, this study does not support the hypothesis that the deletion allele (D) of the ACE gene could be a significant risk factor for VTE, being protective in men.

[Lowering high levels of fasting total homocysteine with folic acid and vitamins B in patients with venous thromboembolism: relationship between response and the C677T methylenetetrahydrofolate reductase (MTHRF) genotype]. / Reducción de concentraciones elevadas de homocisteína con ácido fólico y vitaminas B en pacientes con tromboembolia venosa: relación entre respuesta y genotipo C677T de la metilén tetrahidrofólico reductasa (MTHFR).

BACKGROUND: High levels of plasma total homocysteine (tHcy) are involved in arterial or venous occlusive diseases. It essentially depends on the nutritional status of folic acid (FA) and vitamins B12 or B6, but also on the methylenetetrahydrofolate reductase (MTHFR) enzymatic activity. We aim to evaluate the response of the hyperhomocysteinemia (HHcy) to a standard schedule of vitamin supplementation, according with the MTHFR genotype. PATIENTS AND METHODS: 227 patients, diagnosed with venous thromboembolism (VTE) were analysed for tHcy (in fasting conditions), and for the MTHFR-C677T gene polymorphism. When the tHcy exceeded the cut-off point (men = 16, women = 15 mumol/l), the patients were supplemented with a dose equivalent to 1 mg FA, 0.2 mg B12 and 100 mg of B6, daily by 6 weeks. Afterwards they were reanalysed and the reduction was stratified by MTHFR genotype, looking for any difference in the response. RESULTS: The mean fasting tHcy was 12.3 mumol/l (SD = 8). The 51 hyperhomocysteinemic patients (22%) were older (65.1 y) than the normal ones (55.0 y) (p = 0.0001). The treatment was carried out properly in 46 patients (90%). The pre-treatment mean Hcy was 23.2 (SD = 10.5) mumol/l, and it was reduced to 13.0 (SD = 5.9) (p = 0.0001) (mean reduction = 42.1%). By genotype, the C/C reduced from 21.0 to 13.2 mumol/l (37%) (n = 18), the C/T from 25.0 to 12.6 mumol/l (46%) (n = 24), and the abnormal homozygotes T/T from 22.7 to 14.5 mumol/l (39%) (n = 4), although no statistical significant differences were found. In 80% of cases (37/46), tHcy values normalised. A negative correlation (r = -0.471) (p = 0.005) was observed between age and response. CONCLUSIONS: The FA/B6/B12 based therapy reduces in a simple, quick and effective way (> 40% in 6 weeks) the pathologic tHcy levels on a VTE population and this is not influenced by the MTHFR genotype. As HHcy seems related with recurrences of venous thrombosis, we could speculate if it would be useful to analyse routinely the tHcy, attempting reduction in selected cases.

Normal frequencies of the C677T genotypes on the methylenetetrahydrofolate reductase (MTHFR) gene among lymphoproliferative disorders but not in multiple myeloma.

The folate availability seems to be critical for the DNA integrity since it is required for the transfer of methyl groups in the biosynthesis of thymidilate. Although the excessive incorporation of uracils to the DNA can be efficiently removed, this mechanism of reparation produces many double-strand breaks from two opposing nicks. Several chromosomal abnormalities (mainly translocations and deletions perhaps not well understood) are involved in the origin of lymphoproliferative disorders. The TT homozygosity at nucleotide 677 in the gene of methylene tetrahydrofolatereductase (MTHFR), a key enzyme in folate metabolism, was recently linked to a significant protection against colon carcinoma and acute lymphoblastic leukaemia in adults. We analysed the genotype frequencies of C677T-MTHFR in a group of 143 patients with lymphoproliferative disorders (REAL classification) and 200 controls. Overally, the frequencies of the polymorphic allele were similar (35.3% and 32.0% respectively)(P=0.6). We did not find differences between patients and controls except for myeloma/plasmacytoma group (n=26) which showed a CC genotype less than expected (19% vs 46%) (p=0.01) with a frequency ratio of 0.28 (0.10-0.77). Even among the IgG myeloma cases only one patient showed a common genotype (CC) (1/15, 7%) (P=0.003). If these preliminary data are validated with prospective studies, the 677C allele of MTHFR gene could be confirmed as an effective multiple myeloma protective factor (specially for the IgG cases).

The O blood group protects against venous thromboembolism in individuals with the factor V Leiden but not the prothrombin (factor II G20210A) mutation.

We investigated the influence of the ABO blood group in the observed prevalences of the recently described factor V R506Q and factor II G20210A mutations in a thrombotic population. We determined the ABO blood group in a sample of 178 unselected patients (aged 17-83 years), diagnosed at our center with deep vein thrombosis or pulmonary embolism. The results of this study show a high prevalence of thrombosis in the non-O blood group. In the general population, the prevalence as a fraction of the O blood group was 2.69 (confidence interval 1.90-3.82). Of the factor V R506Q carriers (n = 28), only one had O group blood and 27 of 28 were non-O (24 A, one B and two AB). However, within the group of factor II G20210A carriers (n = 17), seven had O, nine A and one B type blood. The prevalence of the factor V R506Q mutation within the O blood group was 2.4% (one of 42), significantly lower than in the A group (23.3%, 24 of 103; P = 0.002), or in the overall non-O group (19.9%, 27 of 136; P = 0.006). This prevalence was similar to that observed previously in the non-thrombotic population in our area (3.5%; P = 0.9). We analyzed the clotting activity of factor VIII and we found higher levels in the non-O group (1.78+/-0.61 U/ml) than in the O blood group (1.30+/-0.51 U/ ml; P < 0.0001). We speculate that factor Va in individuals with the factor V Leiden mutation could interact with the high levels of factor VIII clotting activity as a necessary cofactor.

[20210A mutation of the prothrombin and venous thromboembolism gene]. / Mutacion 20210A del gen de la protrombina y tromboembolismo venoso.

PURPOSE: Various genetic disorders interact with environmental factors to cause thrombotic diseases. Recently, a G to A transition at nucleotide 20210 in the prothrombin gene, has been described in association with venous thromboembolism, in Dutch population. Currently, several reports want to know the frequence of this mutation in other ethnic groups and populations. The aim of this work was to assess the prevalence rates of prothrombin mutation in both, thrombotic and healthy Spanish populations, and to estimate the associated relative risks. We described the clinical features in our series of thrombotic carriers and moreover, we compared a routine clotting test versus DNA analysis in the diagnosis of this anomaly. POPULATION, MATERIAL AND METHODS: The design was a non-matched case-control study. The involved populations were: 187 patients of venous thromboembolic diseases and 200 healthy controls. Patients and controls were genotyped and both, carriers and non-carrier patients, were analyzed by a routine prothrombin clotting assay, to determine the sensibility and specificity and optimal cut off level of the test. RESULTS: The 20210 A allele was identified in 17 patients (9.1%) and in 7 controls (3.5%), with a 2.76-fold increased risk (OR 2.76, 95% CI = 1.12-6.81), in carriers. One patient and none of the controls were homozygous. The clinical characteristics (first manifestation age or thrombotic recurrence) are similar in both, carriers and non-carriers, patient groups. The prothrombin level by a routine coagulometric method was 1.31 +/- 0.14 U/ml (95% CI = 1.24-1.38) for the 20210 A carriers, whereas for the non carrier-patients was significantly lower, 1.06 (95% CI = 1.03-1.08) (p < 0.00001). With a cut off level of 1.25 U/ml, 12/16 (75%) carriers and 14/132 (10.6%) non-carriers were positive. Therefore, the sensibility was 75% and the specificity 89.4%. With a cut off level of 1.40 U/ml the diagnostic efficiency was even worse. CONCLUSIONS: 3.5% of healthy subjects and 9.1% of thromboembolic patients carried this prothrombin mutation with a relative risk of 2.76 (95% CI = 1.12-6.81). The relevant clinical features are similar to the rest of the series. The mean prothrombin level was higher (1.31 U/ml) than in the normal patients (1.06 U/ml), but the clotting test seems inappropriate for a diagnostic purpose.

A patient homozygous for mutation 20210A in the prothrombin gene with venous thrombosis and transient ischemic attacks of thrombotic origin.

It is well established that genetic disorders interact with environmental factors to cause thrombotic diseases. Therefore, antithrombin, protein C, protein S deficiencies and the more recently described factor V Leiden and prothrombin mutations are currently been investigated to explain some thrombophilic states. We report the case of a 63-year-old man who developed two transient ischemic attacks and two years later an extensive femoro-iliac venous thrombosis. He was genotyped as FV R506Q negative and FII G20210A positive in homozygous state (FII 20210AA).