Research Article

Higher Glycemic Exposure at the Time of Diagnosis of Diabetes after Kidney Transplant may be Associated with Reduced Patient Survival

Peng Chin Kek1,*, Hong Chang Tan1, Terence Yi Shern Kee2, Su-Yen Goh1, Yong Mong Bee1
1Department of Endocrinology, Singapore General Hospital, Singapore
2Department of Renal Medicine, Singapore General Hospital, Singapore
*Corresponding author:

Peng Chin Kek, Department of Endocrinology, Singapore General Hospital, Outram Road, Singapore 169608, Telephone: 65-63214654, Fax: 65-62273576, Email: kek.peng.chin@sgh.com.sg

Keywords:

Diabetes mellitus, Glycemic exposure, Kidney transplantation, New onset diabetes

Post-transplant diabetes mellitus (PTDM) is a serious metabolic complication of kidney transplantation that is associated with higher rates of graft failure, cardiovascular disease, and death. The aim of this study was to determine the association between higher HbA1c at diagnosis of the PTDM with long-term graft and patient survival after kidney transplantation. A retrospective review of solitary kidney recipients transplanted was done. Of the 388 eligible patients, PTDM was reported in 94 participants (24.2%). HbA1c was not available in 33 and were excluded from analysis. Mean HbA1c and glucose level at diagnosis were 9.0 ± 3.1% and 20.9 ± 15.1 mmol/l. Of the 61 participants analyzed, 33 (54.1%) had HbA1c ≥ 8.0%. Participants with HbA1c ≥ 8.0% had poorer long-term survival (seven deaths) using Kaplan-Meier analysis (p=0.041) compared to those with HbA1c < 8.0% (one death). Overall survival was 97.0%, 78.2%, and 54.1% at one, five, and 10 years post-transplant for those with HbA1c ≥ 8.0% in comparison to 100.0%, 96.2%, and 96.2% for those with HbA1c < 8.0%. No difference in graft survival was noted.

Our data suggest that the higher degree of glycemic exposure at the onset of diabetes may be associated with a poorer long-term survival. Monitoring of glucose levels post-transplantation is important to avoid high glycemic exposure.

PTDM: Post-Transplant Diabetes, USRDS: United States Renal Data System, UKPDS: United Kingdom Prospective Diabetes Study

Post-transplant diabetes (PTDM) is a common problem after renal transplant. A meta-analysis of observational studies and randomized controlled trials show that the cumulative incidence of PTDM at one year post-transplantation varies from 2 to 50% [1]. This wide estimate is largely due to the various definitions of PTDM used in the earlier studies. In recent years, data from the United States Renal Data System (USRDS) studies demonstrate an augmented PTDM incidence of 14–16% in the first transplant year after kidney transplant, declining thereafter to an annual incidence of 4–6% similar to pre-transplantation baseline. The cumulative incidence is 24% at three years after transplantation [2,3]. We have previously reported cumulative incidences of PTDM after one, three, and five years post-renal transplantation as 15.8%, 22.8%, and 24.5% respectively [4]. PTDM is a serious metabolic complication of kidney transplantation that predisposes patients to graft dysfunction, cardiovascular disease, and death.1, 2 Data from USRDS demonstrates that PTDM is associated with an 87% increased risk of mortality (P < 0.001), a 63% increased risk of graft failure (P < 0.0001), and a 46% increased risk of death-censored graft failure (P < 0.0001) compared of patients without PTDM [3]. This is supported by our earlier report in the Singaporean population [4]. Although PTDM has been associated with greater mortality and morbidity, the effect of the magnitude/severity glycemic exposure at the time of diagnosis on graft survival and mortality is unclear. A strong association between greater glycemic exposures with mortality has been demonstrated in the United Kingdom Prospective Diabetes Study (UKPDS). A higher HbA1C among individuals with newly diagnosed T2DM in the UKPDS was associated with higher all-cause mortality [5]. Thus, we believe that similar effects are seen in the PTDM population and higher glycemic exposure at the time of diagnosis, as indicated by a greater HbA1C, would adversely affect graft survival and mortality. In this study, we aim to study the effect of greater of glycemic exposure at diagnosis of PTDM on long-term graft and patient survival after kidney transplantation in our center.

This was a single center retrospective review of solitary kidney allograft recipients transplanted in our hospital between 1 January 1998 and 31 December 2007. A total of 432 patients were transplanted during this 10-year period of which 388 were eligible for this study. Reasons for exclusion were pre-existing diabetes mellitus (n=16), death within the first two weeks after transplantation (n=3), graft loss within the first two weeks after transplantation (n=10), and follow-up duration of less than six months (n=15). Data were collected retrospectively from transplant charts and electronic medical records (EMR) according to institutional ethical guidelines. Pre-transplant information obtained from the participants included age at transplantation, gender, ethnicity, height, weight, calculated body mass index (BMI), time in renal replacement therapy, type of renal replacement therapy, hepatitis C status at time of transplant, human leukocyte antigen (HLA) phenotype, total number of HLA mismatches, donor type, age, and gender. Post-transplant information included type of immunosuppressive drugs that was started after transplantation (intention-to-treat), acute rejections, cytomegalovirus infection status, duration of follow-up, and duration of graft and patient survival.

A diagnosis of PTDM was made in patients who after the first two weeks post-transplant had at least two abnormal glucose levels i.e. random plasma glucose (RPG) levels ≥ 11.1 mmol/L [200 mg/dl] and/or fasting plasma glucose (FPG) levels ≥ 7.0 mmol/L [126 mg/dl]) taken on separate occasions, as published in the American Diabetes Association (ADA) guidelines. Chart and EMR review of all available plasma glucose levels were completed for the purpose of diagnosing PTDM. The types of treatment administered for PTDM (at diagnosis and six months later) and the HbA1c measured upon the diagnosis of PTDM were also recorded. PTDM was reported in 94 of the 388 participants (24.2%). HbA1c on diagnosis of PTDM was not available in 33 participants while the remaining was included in the analysis. Higher glucose exposure was defined as HbA1c ≥ 8.0%. This level was selected as it was the suggested less stringent cut-off for glycemic target by ADA [6] and in the ACCORD study, more death were reported in subjects with baseline HbA1c of > 8%7. Participants were subsequently divided into two groups according to their HbA1c levels [5].

All participants received corticosteroids, beginning with a single pre-operative intravenous bolus of 1 g hydrocortisone, followed by 30 mg/day of prednisolone per oral post-transplant, with gradual tapering to 4–6 mg/day prednisolone by the third month. Besides corticosteroids, standard immunosuppression consisted of a calcineurin inhibitor (e.g. cyclosporine or tacrolimus) or mammalian target of rapamycin (mTOR) inhibitor (e.g. sirolimus) and anti-proliferative agents (azathioprine or mycophenolate mofetil). The local protocol for treating clinically suspected or biopsy proven acute rejection was three pulses of 500 mg intravenous methylprednisolone [6,7].

Statistical analysis

Results were expressed as mean standard deviation or median and range. Categorical variables were compared using Chi-squared test. Data was first examined for normality and the Students t-test was used examine between group differences for variables with parametric distribution while the Mann-Whitney U-test was used for variables with non-parametric distribution. Graft and patient survival were calculated using Kaplan-Meier survival curves after censoring for death and graft loss respectively and log rank test was used to compare survival curves. Results were considered statistically significant for p <0.05. All statistical analyses were carried out using SPSS for Windows version 18.0 (SPSS Inc., Chicago, IL, USA).

Participant characteristics

Sixty-one kidney allograft participants with PTDM and with HbA1c recorded on diagnosis were included in the analysis. The mean age at transplantation was 46.9 years and 52.5% of the participants were female. The percentages of participants in the three major ethnic groups were as follows: 82.0% Chinese, 14.8% Malay, and 3.3% Indian. Of the 61 participants, 83.6% received their kidney allograft from a deceased donor. The median time in renal replacement therapy was 89 months (range: 4.4–480.7), with 96.7% of participants receiving hemodialysis. Median follow-up duration was 48.2 months (range 3.5–127.2). Median time to diagnosis of the PTDM was 7.4 months (range 0.5–112.5), with 50.8% diagnosed within the first six months. Mean HbA1c level on diagnosis was 9.0 ± 3.1% and glucose level on diagnoses of PTDM was 20.9 ± 15.1%.

Participants’ characteristics in the groups according to HbA1c

Table 1 compared participants with HbA1c < 8% and those with HbA1c ≥ 8%. There were no significant differences in the distribution of gender, weight, and BMI between the two groups. However, participants with HbA1c ≥ 8% tended to be slightly older with a mean age of 48.3 ± 6.0 years compared to 45.3 ± 9.2 years in those with HbA1c < 8% (p=0.122). They were diagnosed with PTDM earlier, with median onset of PTDM at 4.4 ± 26.7 months compared to 8.1 ± 26.2 months although not statistically significant. Mean HbA1c level was 11.2 ± 2.5% in the HbA1c ≥ 8% group compared to 6.4 ± 0.8% in the group with HbA1c < 8% (p<0.001). The HbA1c levels became statistically not significant between the two groups at first and second year after diagnosis, although the levels were still slightly higher for those from the HbA1c ≥ 8% group, 7.0 ± 1.7% versus 6.25 ± 1.0% (p=0.056) and 6.7 ± 1.2% versus 6.4 ± 1.2% (p=0.455), respectively. Mean glucose levels at diagnosis of PTDM were 14.1 ± 4.9 and 26.7 ± 18.3 mmol/L for the HbA1c < 8% and HbA1c ≥ 8% groups respectively. No significant difference was noted in the creatinine level at diagnosis and at year 1.

 

HbA1c <8% (n=28)

HbA1c ≥8% (n=33)

p value

Numbers

28 (45.9%)

33 (54.1%)

 

Age at transplant (years)

45.3 ± 9.2

 

48.3 ± 6.0

0.122

Gender (% male)

57.1

39.4

0.167

Ethnicity

Chinese

22 (78.6%)

28 (84.9%)

0.809

Malay

5 (17.9%)

4 (12.1%)

 

Indian

1 (3.5%)

1 (3.0%)

 

Weight (kg)

61.4 ± 13.2

56.5 ± 11.4

0.132

BMI (kg/m2)

23.1 ± 4.7

22.5 ± 4.1

0.614

Pre-transplant characteristics

Renal replacement therapy (%)

 

Hemodialysis

96.4

97.0

0.906

Peritoneal dialysis

3.6

3.0

 

Median duration of renal replacement therapy (months)

 

61.2 ± 91.4

91.9 ± 47.1

0.807

Type of transplant (%)

Living

21.4

12.1

0.328

Deceased

78.6

87.9

 

Donor age (years)

40.2 ± 12.0

41.3 ± 12.4

0.734

Gender of donor

(male %)

60.7

66.7

0.515

Immunosuppressants (%)

Prednisolone

96.4

100

0.274

Cyclosporine A

82.1

81.8

0.974

Tacrolimus

14.3

15.2

0.924

Sirolimus

10.7

18.2

0.412

Azathioprine

35.7

36.4

0.958

MMF

53.6

48.5

0.692

Post-transplant characteristics

Median onset of PTDM (months)

8.1 ± 26.2

4.4 ± 26.7

0.913

Glucose level at diagnosis (mmol/L)

14.1 ± 4.9

26.7 ± 18.3

0.001

HbA1c level at diagnosis

6.4 ± 0.8

11.2 ± 2.5

0.000

HbA1c level at year 1

6.25 ± 1.0

7.0 ± 1.7

0.056

HbA1c level at year 2

6.4 ± 1.2

6.7 ± 1.2

0.455

Creatinine level at diagnosis (umol/l)

182.1 ± 93.2

188.4 ± 99.4

0.597

Creatinine level at year 1

165.0 ± 80.5

176.9 ± 103.9

0.351

GFR (ml/min) at diagnosis

45.3 ± 20.0

48.2 ± 24.6

0.471

GFR at year 1

49.5 ± 21.0

50.3 ± 23.1

0.792

Treatment of PTDM

Treatment for PTDM at diagnosis (%)

50

90.9

0.000

Treatment with oral hypoglycemic agents

46.4

57.6

0.385

Treatment with Insulin

3.6

33.3

0.004

Treatment for PTDM at 6 months (%)

35.7

84.8

0.000

Treatment with oral hypoglycemic agents

28.6

60.6

0.008

Treatment with Insulin

7.1

67.9

0.017

Acute rejection

11 (39.3%)

15 (45.5%)

0.627

Graft failure

6 (21.4%)

5 (15.2%)

0.525

Median graft survival (months)

45.8 ± 40.5

46.0 ± 39.2

0.933

Death

1 (3.6%)

7 (21.2%)

0.042

Median follow up duration (months)

61.9 ± 40.5

46.0 ± 39.2

0.633

 

Table 1. Characteristics of participants with HbA1c < 8% and those with HbA1c ≥ 8%.

The modes of renal replacement were similar between the two groups but the group with HbA1c ≥ 8% had a longer duration of renal replacement therapy prior to the kidney transplants with a median duration of 91.9 ± 47.1 months compared to 61.2 ± 91.4 months (p=0.807). No significant difference was noted for donor type and immunosuppressant usage. Treatment was required for 50% of patients on diagnosis of patients with HbA1c < 8% and this was reduced to 35.7% at six months. Those with HbA1c ≥ 8% required treatment in 90.9% and 84.8% at diagnosis and at six months, respectively.

There were more episodes of acute rejection (15 events) in the group with HbA1c ≥ 8% compared to the group with HbA1c < 8% (11 events) but this was not statistically significant (p=0.627). There were, however, no significant differences in the episodes of graft failure in the two groups with six events in the participants with HbA1c < 8% and five events in the participants with HbA1c ≥ 8% (p=0.042).

Overall graft survival (censored for death) was not statistically different between the two groups (p=0.702) (Figure1 and Figure 2). Subjects with HbA1c ≥ 8.0% had poorer long-term survival (seven deaths) using Kaplan-Meier analysis (p=0.041) compared to those with HbA1c < 8.0% (one death). Overall patient survival was 97.0%, 78.2%, and 54.1% at one, five, and 10 years post-transplant for those with HbA1c ≥ 8.0% and 100.0%, 96.2%, and 96.2% for those with HbA1c < 8.0% at one, five, and 10 years post-transplant.

clyto access
Figure 1.Kaplen-Meier analysis of overall graft survival (censored for death) in participants with HbA1c < 8% and those with HbA1c ≥ 8%, showing no statistically different between the two groups (p=0.702).

clyto access
Figure 2.Kaplan-Meier analysis showing subjects with HbA1c ≥ 8.0% having poorer long-term survival in comparison to those with HbA1c < 8.0% (p=0.041).

There was a tendency towards higher death events in participants with higher glycemic exposure. There were 3.6% of death events in the group with HbA1c < 8%, 20% in those with HbA1c between 8–12%, and 25% in those with HbA1c ≥ 12% (Figure 3), although this was not statistically significant (p=0.118). Causes of death included: acute myocardial infarction (n=2), pneumonia (n=4), and urinary tract infection (n=1). One patient passed away out of hospital and thus the cause of death was not recorded but this patient was admitted for dissemination cytomegalovirus infection prior to death.

clyto access
Figure 3.Percentage of death events in groups with HbA1c < 8%, HbA1c between 8–12%, and those with HbA1c ≥ 12%.

Diabetes mellitus is a common complication after kidney transplantation. Studies using uniform criteria for diagnosing PTDM according to ADA guidelines showed that the cumulative incidence of PTDM at one year ranged from 7.0%–19.0% [8-11]. Data from the USRDS studies demonstrated an augmented NODM incidence of 14–16% in the first transplant year and 24% at three years after transplantation 2, 3 We have previously reported similar cumulative incidences of PTDM after one, three, and five years post-transplantation which were 15.8%, 22.8%, and 24.5% respectively [4]. PTDM was particularly accelerated in the first few post-transplant months with 50.8% diagnosed within the first six months in our cohort. This is consistent with other published reports [10,12,13].

The importance of PTDM cannot be over-emphasized. Cosio et al analyzed the data from 1811 adult renal allograft recipients transplanted between 1983 and 1998 and showed that PTDM is an independent predictor of reduced patient survival (HR 1.80, 95% CI 1.35 to 2.41, P = 0.001) [14]. Data from USRDS demonstrated that in comparison to patients with no diabetes, PTDM was associated with an 87% increased risk of mortality (P < 0.001) [3]. PTDM has also been reported to 8 reduce graft survival. In particular, Kasiske et al reported a 63% increased risk of graft failure (P < 0.0001) and 46% increased risk of death-censored graft failure (P < 0.0001) in patients with PTDM [3]. We previously reported that PTDM was associated with significant reduction in both patient and graft survival [4]. However, data on the degree of glycemic exposure on long-term patient and graft survival was very limited in the transplanted population.

Previous studies in patients with type 2 diabetes in the general population found an association between the degree of hyperglycemia, increased risk of cardiovascular mortality, and all-cause mortality [15-17]. Data from the UKPDS has suggested an association between glycemia and mortality related to diabetes, and all-cause mortality.5 Each percent decrease in HbA1c was associated with a 21% decrease in death related to diabetes (p<0.001) and a 14% decrease in all-cause mortality (p<0.001). The same observation was reported in another prospective trial which found a continuous relationship between HbA1c and all-cause mortality [18]. Each percent increase in HbA1c was associated with a relative risk of death from any cause of 1.24 (95% CI, 1.14 to 1.34, p<0.001) in men and a relative risk of 1.28 (95% CI, 1.06 to 1.32; p<0.001) in women. Interestingly in the ACCORD trial, those with baseline HbA1c of > 8% had more death from any cause in comparison to those with HbA1c ≤ 8% (256 events vs 204 events) [7].

One question is whether we can extrapolate these associations to patients with PTDM. One study looked at the effects of post-transplant glycaemia on long-term survival [19]. In this study, 1410 consecutive renal transplant recipients without known diabetes underwent an oral glucose challenge test at 10 weeks after their transplant and were observed for a median of 6.7 years (range 0.3–13.8). Recipients with PTDM had higher all-cause and cardiovascular mortality (HR: 1.54. 95% CI 1.09 to 2.17 and 1.80, 95% CI 1.10 to 2.96) after adjusting for traditional risk factors. Each 1 mmol/l increase in the fasting glucose or two hour plasma glucose was associated with 11% increase in all-cause mortality and 6% increase in cardiovascular mortality. However, the degree of glycemic exposure as measured by HbA1c was not reported in this study. Nevertheless, this observation might still be relevant in suggesting that a higher degree of glycemic exposure may be associated with poorer long-term survival. Our study supported this association. We demonstrated a significant poorer long-term survival in participants with HbA1c ≥ 8% on 9 diagnosis of PTDM. This was despite similar HbA1c levels at years one and two after transplantation. There was a tendency towards higher death events in subjects with higher glycemic exposure with 3.6% of death event in the group with HbA1c < 8%, 20% in those with HbA1c between 8–12%, and 25% in those with HbA1c ≥ 12% (Figure 3.). This observation suggests that higher degree of glycemic exposure, particularly on diagnosis may affect the long-term survival. The degree of glycemic exposure at diagnosis of PTDM did not seem to have any effect on the long-term graft survival.

This study has some limitations that need to be acknowledged. This was a retrospective analysis that is subject to reporting bias or error inherent in registry database, including the non-random assignment of patients to different immune suppression protocols. However, the baseline profiles were similar between the two groups. In addition, pre-existing glucose intolerance was not systematically assessed in this study cohort by performing pre-transplantation oral glucose tolerance test. It is thus possible that some participants may have pre-existing impaired glucose metabolism. Our sample size of 61 was relatively small; not all participants with PTDM had HbA1c measured at diagnosis. The total number of events was also small in our study which may potentially affect the overall result.

In conclusion, this study shows that the degree of glycemic exposure at the diagnosis of PTDM may be associated with poorer long-term survival. More research would need to be carried out to confirm this association in the transplanted population. However, our findings do suggest that it is purulent to monitor glucose closely, and even as part of routine post-transplant protocol, to detect PTDM and high glucose level so that exposure to high degree of glucose can be prevented.

No conflict of interest

This research was approved by the Institutional Review Board of Singapore General Hospital. There was no pharmaceutical funding for this study. We would like to acknowledge Miss Ai Ling Yeo for revision of the manuscript.

1. Montori VM, Basu A, Erwin PJ, Velosa JA, Gabriel SE, et al. (2002) Post transplantation diabetes: a systematic review of the literature. Diabetes care 25: 583-592.
2. Woodward RS, Schnitzler MA, Baty J, Lowell JA, Lopez-Rocafort L, et al. (2003) Incidence and cost of new onset diabetes mellitus among U.S. wait-listed and transplanted renal allograft recipients. Am J transplantation : official J Am Soci of Transplantation and the Am Society of Transplant Surgeons 3: 590-598.
3. Kasiske BL, Snyder JJ, Gilbertson D, Matas AJ (2003) Diabetes mellitus after kidney transplantation in the United States. Am J transplantation : official J Am Soci of Transplantation and the Am Society of Transplant Surgeons 3: 178-85.
4. Bee YM, Tan HC, Tay TL, Kee TY, Goh SY, et al. (2011) Incidence and risk factors for development of new-onset diabetes after kidney transplantation. Annals Academy Med, Singapore 40: 160-168.
5. Stratton IM, Adler AI, Neil HA, Matthews DR, Manley SE, et al. (2000) Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35): prospective observational study. Bmj 321: 405-412.
6. American Diabetes A (2017) 6 Glycemic Targets. Diabetes care 40 (Suppl 1): S48-S56.
7. Action to Control Cardiovascular Risk in Diabetes Study G, Gerstein HC, Miller ME, Byington RP, Goff DC, Bigger JT, et al. (2008) Effects of intensive glucose lowering in type 2 diabetes. The New England J Med 358: 2545-2559.
8. Roland M, Gatault P, Doute C, Buchler M, Al-Najjar A, et al. (2008) Immunosuppressive medications, clinical and metabolic parameters in new-onset diabetes mellitus after kidney transplantation. Transpl Int 21: 523-530.
9. Kiberd M, Panek R, Kiberd BA (2006) New onset diabetes mellitus post-kidney transplantation. Clin Transplant 20: 634-649.
10. Gourishankar S, Jhangri GS, Tonelli M, Wales LH, Cockfield SM (2004) Development of diabetes mellitus following kidney transplantation: a Canadian experience. Am J Transplant 4: 1876-1882.
11. Chien YS, Chen YT, Chuang CH, Cheng YT, Chuang FR, et al. (2008) Incidence and risk factors of new-onset diabetes mellitus after renal transplantation. Transplant Proc 40: 2409-2411.
12. Cosio FG, Pesavento TE, Osei K, Henry ML, Ferguson RM (2001) Post-transplant diabetes mellitus: increasing incidence in renal allograft recipients transplanted in recent years. Kidney Int 59: 732-737.
13. Joss N, Staatz CE, Thomson AH, Jardine AG (2007) Predictors of new onset diabetes after renal transplantation. Clin Transplant 21: 136-143.
14. Cosio FG, Pesavento TE, Kim S, Osei K, Henry M, et al. (2002) Patient survival after renal transplantation: IV. Impact of post-transplant diabetes. Kidney Int 62: 1440-1446.
15. Groeneveld Y, Petri H, Hermans J, Springer MP (1999) Relationship between blood glucose level and mortality in type 2 diabetes mellitus: a systematic review. Diabetic medicine: a journal of the British Diabetic Association 16: 2-13.
16. Uusitupa MI, Niskanen LK, Siitonen O, Voutilainen E, Pyorala K (1993) Ten-year cardiovascular mortality in relation to risk factors and abnormalities in lipoprotein composition in type 2 (non-insulin-dependent) diabetic and non-diabetic subjects. Diabetologia 36: 1175-1184.
17. Wei M, Gaskill SP, Haffner SM, Stern MP (1998) Effects of diabetes and level of glycemia on all-cause and cardiovascular mortality. The San Antonio Heart Study. Diabetes care 21: 1167-1172.
18. Khaw KT, Wareham N, Bingham S, Luben R, Welch A, et al. (2004) Association of hemoglobin A1c with cardiovascular disease and mortality in adults: the European prospective investigation into cancer in Norfolk. Annals internal med 141: 413-420.
19. Valderhaug TG, Hjelmesaeth J, Hartmann A, Roislien J, Bergrem HA, et al. (2011) The association of early post-transplant glucose levels with long-term mortality. Diabetologia 54: 1341-1349.

Citation: Kek PC, Tan HC, Kee TYS, Goh SY, Bee YM (2017) Higher Glycemic Exposure at the Time of Diagnosis of Diabetes after Kidney Transplant may be Associated with Reduced Patient Survival. J Diabetes Care Endocrinol 1:001.

Published: 15 November 2017

Reviewed By : Maurilio de Souza Cazarim,

Copyright:

© 2017 Kek et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.