Comparison of blood pressure values and expression of genes associated with hypertension in children before and after hematopoietic cell transplantation

Hypertension is a well-known late effect of hematopoietic cell transplantation (HCT), but no markers predicting its development are known. Our aim was to assess short-term blood pressure (BP) values and expressions of hypertension-associated genes as possible markers of hypertension in children treated with HCT. We measured systolic blood pressure (SBP) and diastolic blood pressure (DBP), using both office procedure and ambulatory BP monitoring (ABPM) in children before HCT and after a median of 6 months after HCT. We compared the results with two control groups, one of healthy children and another of children with simple obesity. We also performed microarray analysis of hypertension-associated genes in patients treated with HCT and children with obesity. We found no significant differences in SBP and DBP in patients before and after HCT. We found significant differences in expressions of certain genes in patients treated with HCT compared with children with obesity. We concluded that BP values in short-term follow-up after HCT do not seem to be useful predictors of hypertension as a late effect of HCT. However, over expressions of certain hypertension-associated genes might be used as markers of hypertension as a late effect of HCT if this is confirmed in larger long-term studies.


Results
Office blood pressure measurements. Mean office SBP and DBP values were significantly higher in the obesity control group compared with the healthy control group (124/75.7 vs. 111/66.4 mmHg; P < 0.001), while no significant differences were found between the pre-HCT group or post-HCT group and the healthy control group (Table 1).
Mean office SBP and DBP values were significantly higher in the obesity control group compared with both the pre-HCT group (124/75.7 vs. 108/66.9 mmHg; P < 0.001) and post-HCT group (124/75.7 vs. 104/62.8 mmHg; P < 0.001) ( Table 2), while no significant differences were found between the pre-HCT and post-HCT groups ( Table 3, results in paired tests see Supplementary Table S5).
Ambulatory blood pressure measurements. ABPM measurements revealed significantly higher values of mean arterial pressure (MAP) and MAP percentile, mean SBP, and DBP percentile in the obesity control group compared with the healthy control group. SBP percentile and mean DBP values were also higher in this group, but the differences were not significant. Comparisons of the pre-HCT group and the healthy control group revealed significant differences with respect to mean DBP values and MAP percentile. No significant differences were found between the post-HCT and healthy control groups (Table 4).
ABPM measurements revealed significantly higher SBP in the obesity control group compared with the pre-HCT group. Other ABPM parameters were also higher in the obesity control group, but the differences were not significant. Comparisons of the post-HCT group with the obesity control group revealed significantly higher mean SBP and SBP percentile in the obesity control group. Other ABPM parameters were also higher but the differences were not significant (Table 5). No significant differences in the ABPM parameters were found between the pre-HCT and post-HCT groups ( Table 6; results in paired tests see Supplementary Table S6).  www.nature.com/scientificreports/ Microarray analysis. Microarray analysis revealed significant differences in expressions of genes associated with hypertension between the pre-HCT and obesity control group, as well as between the post-HCT group and the obesity control group. In the pre-HCT group we found significantly lower expression of AGTR2, BLK, FLJ32810 and TMEM133 genes, and significantly higher expression of MOV10 and WNK1 genes compared with the obesity control group. In the post-HCT group we found significantly lower expression of AGTR2, FLJ32810, NR3C2 and TMEM133 genes, and significantly higher expression of BAT2D1, MOV10, PIK3CG and WNK1 genes compared with the obesity control group. The MOV10 and WNK 1 genes were overexpressed in both pre-HCT and post-HCT group compared with the obesity control group. We found no significant differences in expressions of the hypertension-associated genes between the pre-HCT and post HCT groups, with only a few genes showing near-significant differences and no consistent pattern (including NR3C2 gene with higher expression in the pre-HCT group and PIK3CG with higher expression in the post-HCT group, both differences non-significant). Statistically significant differences in expressions of the respective genes are presented in Table 7. Gene expression profiles are presented in Fig. 1.

Discussion
In our prospective analysis of patients treated with HCT compared with control groups of healthy children and obese children with no comorbidities we found no significant differences of BP values between the pre-HCT and the post-HCT groups, as well as the healthy controls. This was seen despite the fact that numerous patients have received drugs that may have contributed to hypertension, including glucocorticoids and ciclosporin. Also, the age, sex and disease status of these groups was not associated with a relative increase in the known risk of hypertension in any of these groups. We found elevated mean office BP values in the group of otherwise healthy obese individuals compared with other groups. This was consistent with a generally elevated prevalence of hypertension in this population, which may be over 30%, compared with over 10% in otherwise healthy children and adolescents without obesity [6][7][8] . BP parameters measured using ABPM generally followed a similar pattern with one exception, as mean DBP values and MAP percentile were significantly higher in the pre-HCT group than in healthy controls. Moreover, not all of the differences reached statistical significance.
HCT causes a variety of adverse effects, both early and late, that constitute significant health burden in patients treated with this therapeutic modality, affecting their quality of life and survival 9 . Late effects of HCT may appear over several years or decades after the procedure and the association between their emergence and HCT may not be evident 10 . Hypertension and obesity are very prevalent in contemporary societies [11][12][13][14] . Hypertension is also a well-known late effect of HCT 1-3. Hoffmeister et al. found that overall prevalence of hypertension in long-term survivors after pediatric HCT was 15%, which was up to three times higher than in the reference population 3 . The mechanisms of the emergence of hypertension in post-HCT patients are still not well understood. Contributing factors may include drugs used after the procedure, such as ciclosporin or glucocorticoids, as well as kidney or vascular injury caused by the procedure, however the relative importance of these factors has not been unequivocally proven and there are currently no known markers predicting the development of hypertension in individual patients [15][16][17][18][19][20] . Table 5. ABPM parameters in the pre-HCT group and post-HCT group compared with the obesity control group.

Parameter (n of measurements)
Pre-HCT (53) p  www.nature.com/scientificreports/ Our microarray analysis of the expressions of hypertension-associated genes revealed several significant differences in the gene expressions between the obesity control group, and the pre-HCT and post-HCT groups. A very interesting finding were significantly higher expressions of certain hypertension-associated genes in patients treated with HCT (both before and after the procedure) compared with obese individuals with no comorbidities. We found no significant differences in the expressions of the assessed genes between the pre-HCT and post-HCT groups.
The evidence of the role of genetic mechanisms in the pathogenesis of hypertension is accumulating [21][22][23][24] . This issue was comprehensively reviewed by Ehret and Caulfield, who listed 12 genes associated with monogenic types of hypertension and 43 genetic variants that are associated with multigene pathogenesis of hypertension 5 . In our study we compared expressions of these genes (and a few more new additions) in patients before and after HCT, as well as healthy and obese controls. Comparing patients before HCT with obese controls we found significant overexpression of four genes (AGTR2, BLK, FLJ32810 and TMEM133) in obese individuals, which could have been expected. However, quite unexpectedly, we found significant overexpression of two genes (MOV10 and WNK1) and near-significant overexpression of another gene (SH2B3) in patients before HCT. Similar findings were seen in the comparison of patients after HCT and obese controls, where significant overexpression of four genes (AGTR2, FLJ32810, NR3C2 and TMEM133; three of them overlapping the former comparison) was found in obese subjects, while significant overexpression of another four genes (BAT2D1, MOV10, PIK3CG Table 7. Statistically significant differences of gene expressions between pre-HCT and post-HCT groups and the obesity control group.  Figure 1. Differences in the expression of blood pressure-associated genes between pre-HCT, post-HCT, and obesity control group. Microarray results are shown as a heatmap and include the hypertension-associated genes for which significant differences in expressions between the study groups were found. Colored rectangles represent transcript abundance in the blood. www.nature.com/scientificreports/ and WNK1) was found in patients after HCT. The differences between patients before and after HCT were not significant and no consistent pattern was found. Of the genes overexpressed in patients before or after HCT, WNK1 is associated with a monogenic type of hypertension, while BAT2D1, MOV10, PIK3CG are genetic variants associated with BP 5,25 . MOV10 and WNK 1 genes were overexpressed both before and after HCT compared with the obesity control group. The altered expression of the genes was found in mononuclear cells. These were host cells in patients before HCT, and graft cells in patients after HCT. The differences in gene expression in host and graft cells are known and are subject of research focused on transplant rejection and tolerance [26][27][28][29][30] . The effects of microenvironment on gene expression are also implicated with respect to tumor angiogenesis 31 . This suggests that donor immune and hematopoietic system may influence host tissues and vice versa. This may also include the pathways taking part in regulation of BP. However, to our knowledge no data have been published to date on expression of the genes that may have a role in the development of late effects of cancer treatment. These findings are intriguing, as they might suggest the effect of some host factors that are beyond the direct effects of the graft. Hypothetically, either an underlying condition that was an indication for HCT, a prior treatment (particularly glucocorticoids), an HCT procedure itself, or a combination of these factors, might have some, yet unexplained, effects on expressions of the genes that could play a role in the future development of hypertension. Our study has several limitations. These include low numbers of patients in the study groups, variable age of patients in the study and children in the control groups (caused by the inclusion of individual patients at the extremes of the pediatric age range), heterogenous population of healthy controls, heterogeneous indications for HCT in the study group, difference in the number of patients between pre-HCT and post-HCT groups caused by a high number of patients who died or were lost to follow up due to complications of their underlying diseases or HCT procedure, observational single-center design and relatively short follow-up. Another limitation is lack of comparison of gene expressions between patients before/after HCT and healthy controls. This was originally performed, but for technical reasons we found the results non-reportable, and thus we have not included this data set in our analysis.
In conclusion, our results show that in short-term follow-up BP values in pediatric patients treated with HCT were not significantly different before and after the procedure despite the use of agents potentially implicated in the development of hypertension. No significant differences were found compared with healthy controls, while the BP values were significantly lower than in the obese controls (except for two ABPM parameters). This suggests that short-term BP values after HCT may not be a useful marker of the development of hypertension as a late effect. In an additional analysis we found overexpressions of individual hypertension-associated genes in patients before and after HCT compared with obese controls. This is an intriguing finding, as it might suggest a role of some genetic factors in the development of hypertension in patients treated with HCT. This is a preliminary research and has numerous limitations reviewed above in the discussion, but in our opinion it merits further research to investigate the possible role of altered expression of hypertension-associated genes in the development of hypertension in this group of patients. If this is confirmed in larger studies, such genes might be used as markers predicting long-term risk of hypertension in patients treated with HCT.

Methods
We prospectively compared four groups of children.
The group of patients assessed before HCT (pre-HCT group) included 44 patients (31 boys, 13 girls) aged 1.5-19 years (median 9.9 years), consecutively referred to the Stem Cell Transplantation Centre of the University Children's Hospital in Krakow. Characteristics of the pre-HCT group are presented in Supplementary Table S1. Indications for HCT are presented in Supplementary Table S2, and types of transplantation procedures in Supplementary Table S3. Patients with malignancies were referred for HCT in complete remission (leukemia) or at least in a very good partial remission (solid tumors). HCT procedures were performed from June 2009 to August 2012. Before HCT all patients underwent a standard screening including detailed assessment of cardiac function (echocardiography) to exclude cardiomyopathy or other cardiovascular abnormalities. None of the patients had either cardiovascular disease, or a prior history or diagnosis of hypertension. None of the patients received antihypertensive treatment before and/or after HCT.
The group of patients assessed after HCT (post-HCT group) included 27 children (20 boys, 7 girls), aged 2.8-19.5 years (median 11.2 years) from the pre-HCT group in whom repeated investigations were performed after a median of 6 months after HCT. Characteristics of the post-HCT group are presented in Supplementary  Table S4. The 6-month period of analysis was chosen because corticosteroids and immunosuppressive therapy are usually discontinued earlier than 6 months after HCT and therefore they would not affect BP results. The post-HCT group did not include 17 children from the pre-HCT group, of whom 12 were lost to follow up before reaching the time point of 6 months after HCT and 5 children died before they could be included in the post HCT group (the causes of death were complications of HCT or disease progression). Systemic glucocorticoids were used in 25 children in the post-HCT group to treat complications of HCT. Moreover, they received other immunosuppressive agents, including tacrolimus, mycophenolate mofetil, ciclosporin and etanercept. During the second follow up assessment five children still received tapered doses of immunosuppressive agents other than glucocorticoids (including ciclosporin).
Two control groups were recruited. The healthy control group included 26 children (11 boys, 15 girls, age range 4.3-16.0 years, median age 12.2 years). These were family donors, siblings of HCT patients or other healthy children unrelated to the patients. They had no acute or chronic diseases currently or in the past and had laboratory test results (complete blood count, and serum alanine aminotransferase and creatinine levels), as well as body mass index values (BMI: mean 20 Microarray analysis. Methodology was described in detail in our previous study 33 . At the time of obtaining the samples the patients have not received glucocorticoids. Blood samples were collected from the pre-HCT, post-HCT and obesity control groups, mononuclear cells were separated, total RNA extraction was performed, and microarray analysis was performed using the GeneChip Human Gene 1.0 ST Array (Affymetrix, Santa Clara, CA, USA). Initial processing of the microarray data was performed using GeneChip Operating Software. DTT data were then transferred using the Transfer Tool software (Affymetrix, Santa Clara, CA, USA). Chip quality was assessed according to the guidelines by Affymetrix. Raw data were processed using the modelbased expression index implemented in dChip (Cheng Li Lab, Ver 2/25/09+). After background subtraction, the data were normalized using quintile normalization. The signal was taken as the measure of mRNA abundance derived from the level of gene expression. Characteristics of the studied genes related to hypertension on the basis of Ehret and Caulfield 5 and EntrezGene database are presented in Supplementary Table S5.
Statistical analysis. All the continuous variables are presented as mean and standard deviation (SD) values. The categorical variables are presented as frequencies and percentages. The Shapiro-Wilk test was used to test for the normality of the data. To examine differences between the independent groups (normally distributed data) Student's t-test or ANOVA (Analysis of Variance) were used. The Mann-Whitney/Kruskal-Wallis tests for non-normally distributed variables were used. The power of the microarray study was estimated using the R pwr package (version 1.2-0). Cohen's d effect size (1.01) was calculated using average within-group SD 0.99 (for log2 values), assuming two-fold as expected change of fold. Significance levels of differences between the experimental groups were calculated for each microarray probe set using a t-test. P-values below 0.05 were considered statistically significant. The statistical analyses were carried out using R 3.0 software (Bioconductor).

Data availability
Data generated or analyzed during this study are included in this published article (and its Supplementary Information files). The datasets generated for this study can be found in the GEO Series accession number GSE88852 (http:// www. ncbi. nlm. nih. gov/ geo/ query/ acc. cgi? acc= GSE88 852).