- Email: [email protected]
Occurrence, outcomes and predictors of portal hypertension in cystic fibrosis: A longitudinal prospective birth cohort study
ARTICLE IN PRESS
JID: JCF
[m5G;October 29, 2019;19:29]
Journal of Cystic Fibrosis xxx (xxxx) xxx
Contents lists available at ScienceDirect
Journal of Cystic Fibrosis journal homepage: www.elsevier.com/locate/jcf
Occurrence, outcomes and predictors of portal hypertension in cystic fibrosis: A longitudinal prospective birth cohort study Cipolli Marco a, Fethney Judith b, Waters Donna b, Zanolla Luisa c, Meneghelli Ilaria a, Shoma Dutt d, Assael Baroukh Maurice e, Gaskin Kevin John d,∗ a
Cystic Fibrosis Center, Azienda Ospedaliera Universitaria Integrata, Verona, Italy Susan Wakil School of Nursing and Midwifery, University of Sydney, Sydney, Australia Cardiology Department, Azienda Ospedaliera Universitaria Integrata, Verona, Italy d Department of Gastroenterology/James Fairfax Institute of Paediatric Nutrition, The Children’s Hospital at Westmead, Sydney, Australia e Scientific Advisor, Adult Cystic Fibrosis Center, University of Milano Medical School, Milano, Italy b c
a r t i c l e
i n f o
Article history: Received 4 April 2019 Revised 29 August 2019 Accepted 24 September 2019 Available online xxx Keywords: Portal hypertension Liver enzymes Incidence/Prevalence
a b s t r a c t Background: The reported prevalence of portal hypertension (PH) in Cystic Fibrosis is variable, incidence rates rarely provided and the utility of liver function tests (LFT’s) early in life to predict PH is questionable. The aims were to (1) determine PH prevalence (P) and incidence rate (IR) and combined mortality transplant (MTX) data in PH vs non-PH patients and (2) to assess association of LFTs in early life with liver disease and PH. Method: (1) A double centre longitudinal cohort study of 577 CF patients diagnosed by newborn screening (NBS) with annual examinations for PH up to 18.5 years of age (max) was performed over 28 years for P, IR, and MTX data; (2) Cox proportional hazard models were used to assess the association of elevated LFTs on 2 or more occasions over 0–6.5 years and PH. Results: 51/577(8.8%) developed PH with an average IR of near 3/10 0 0 patient years per 5 year interval representing young, mid and late childhood respectively in patients 3–18 years of age. Combined mortality/liver transplant occurred in 12/51 (23.5%) PH and 25/526 (4.8%) non-PH (p < 0.001). Elevated enzymes particularly GGT (HR:5.71, 95% CI 3.11–10.47); ALT/GGT (HR: 5.56, 95% CI 2.82–10.98); and ALP/GGT (HR: 5.74, 95% CI 2.78–11.86) were associated with the onset of PH. Conclusion: This birth cohort with annual examination for PH provides an accurate assessment of the prevalence, and IR of PH and MTX of PH vs non-PH. Early elevated LFTs are associated with onset of MBC/PH. © 2019 European Cystic Fibrosis Society. Published by Elsevier B.V. All rights reserved.
Introduction The severe form of CF liver disease, multilobular biliary cirrhosis (MBC) is associated with the almost invariable development of portal hypertension (PH) and the complications of variceal bleeding [1–4] and unusually hepatic encephalopathy [5] or liver fail-
Abbreviations: ALP, alkaline phosphatase; ALT, alanine aminotransferase; CF, cystic fibrosis; GGT, gamma glutamyl transpeptidase; HR, hazard ratio; LFTs, liver function tests; MBC, multilobular biliary cirrhosis; MTx, mortality/transplant; PH, portal hypertension. ∗ Corresponding author at: The Children’s Hospital at Westmead, Locked Bag 4001, Westmead, NSW 2145, Australia. E-mail addresses: [email protected] (C. Marco), [email protected] (F. Judith), [email protected] (W. Donna), [email protected] (Z. Luisa), [email protected] (M. Ilaria), [email protected] (S. Dutt), [email protected] (A. Baroukh Maurice), [email protected] (G. Kevin John).
ure requiring transplantation [6]. These features in combination with splenomegaly and laboratory evidence of hypersplenism facilitate a diagnosis of MBC/PH which can be confirmed with noninvasive ultrasonography [2,7–9] and magnetic resonance imaging. Although the average prevalence in large studies is 5–6% [1,3,9– 16] extreme values of 1% [10,11] to 22% [12,13] have been reported. The latter could represent a marked heterogeneity of MBC/PH but they are likely related to the different study designs/methodology or the age groups being studied. The 1% value was from a postal questionnaire [10] and was subsequently considered an underestimation related to the accuracy of data recorded and collected [1]. Regarding the extreme 2 values near 20%, one was an autopsy study [12] and the other was from a transplant centre [13], with the prevalence calculated from the smaller group undergoing autopsy or being considered for transplant rather than the true population of origin. A more reliable estimate could be obtained from
https://doi.org/10.1016/j.jcf.2019.09.016 1569-1993/© 2019 European Cystic Fibrosis Society. Published by Elsevier B.V. All rights reserved.
Please cite this article as: C. Marco, F. Judith and W. Donna et al., Occurrence, outcomes and predictors of portal hypertension in cystic fibrosis: A longitudinal prospective birth cohort study, Journal of Cystic Fibrosis, https://doi.org/10.1016/j.jcf.2019.09.016
JID: JCF 2
ARTICLE IN PRESS
[m5G;October 29, 2019;19:29]
C. Marco, F. Judith and W. Donna et al. / Journal of Cystic Fibrosis xxx (xxxx) xxx
clinics utilising newborn screening for CF diagnosis and the birth cohort number as the true population of origin. Several publications [1,3,17–19] have reported on the futility of using the common serum liver enzyme tests (LFTs) alanine aminotransferase (ALT), gamma glutamyl transpeptidase (GGT) and alkaline phosphatase (ALP) to predict the occurrence of PH in CF. However, another report [15] suggested that if any 2 enzymes were elevated on 2 or more occasions in the first 5 years of life this scenario was over 6x more common in those with PH than in those without PH in later childhood, but no statistical evaluation was provided. LFTs therefore require further evaluation to assess their utility in predicting the later occurrence/outcomes of PH and liver disease. A collaborative database was developed from 2 of the original longterm neonatal screening clinics, Verona CF centre (VCFC), Italy and the Children’s Hospital at Westmead (CHW) CF Clinic, Australia and contains annual data from birth on both examination signs of MBC/PH and LFTs from 1986–2014.The aims of this study were first to provide reliable prevalence and incidence rates of PH and outcomes (mortality/ liver transplant rates) in comparison with the non-PH sub-group and secondly to determine whether elevated serum LFTs on 2 or more occasions in the first 6.5 years of life are associated with the development of MBC/PH and complications. Methods Study population A longitudinal cohort study was conducted on all CF patients diagnosed by newborn screening (NBS) and referred to VCFC and CHW from January 1st 1986 to January 1st 2007. Patients were followed to a maximum 18.5 years of age or January 2014, date of death or liver transplant whichever was sooner. Genotypically, the centres differed (VCFC < 30% DF508 homozygotes, 50% DF508/other; CHW > 50% DF508 homozygotes, 43% DF508/other) noting the proportions with severe type 1-111 mutations and the mild type1V and V mutations were very similar as were the proportions with pancreatic insufficiency (PI) sufficiency (PS) and neonatal meconium ileus at both centres [20–21]. All patients had confirmatory sweat chlorides with compound heterozygotes having extended mutation analyses as available in each centre. MBC/PH was defined clinically at annual examinations as the presence of a palpable firm multilobulated liver (>90% of cases) and splenomegaly (100%). All PH cases were confirmed initially by ultrasound regarding the presence of a multilobulated liver and splenomegaly and the variable presence of intraabdominal varices and spontaneous porto-systemic shunts. No cases had portal vein thrombosis nor portal cavernomas. MRI angiography was undertaken in less than 10% where the portal vein could not be adequately visualised on ultrasound examination. Endoscopic examination for oesophageal/gastric varices was used rarely to confirm imaging findings but was used in every case with haematemesis (3 cases occurring within the first 3 years after the onset of PH in the 8–12 age range and in 4 cases occurring in 13–16 years of age whose PH was diagnosed 11–12 years of age). Other causes of cirrhosis such as infectious and autoimmune hepatitis, alpha1-antitrypsin deficiency and Wilsons disease were excluded. MTx was the combined mortality and transplants at both centres as numbers in the PH subgroup were low (7 liver Tx, 1 lung Tx and 4 deaths). To evaluate the 3 liver enzymes (ALT, GGT and ALP) as potential early predictors of MBC/PH, the LFTs were collected at diagnosis and annually, with enzyme data censored at 6.0 years of age (allowing a max of 6.5 years for delays in annual exams) or until the time of death or onset of PH if occurring pre-6.5 years. This timepoint was chosen due to (a) previous reporting, but without
statistical analysis, of LFT elevation 0–5 years of life and the later occurrence of CFLD [15] and (b) 6.0 years of age being approximately half the average age of the peak occurrence of MBC [29]. During this interval there were 31/577 (5.4%) patients with no LFT data. Of the remaining 546 patients, 512 had LFTs measured on 2 or more of their annual visits. Reference values for normal children were derived from pooled samples at varying ages with cross referencing to the paediatric literature and elevated values >4 weeks of age defined as above the ULN: ALT 50 IU/L, GGT 25 IU/L and ALP 350 IU/L [22–23]. Statistical analysis Prevalence of PH was (1) the proportion of patients who had developed PH by 1st January 2014 for the total population of 577 patients; (2) the enzyme cohort of 512 patients; and (3) the corresponding data for the 2 clinics separately. Incidence rates (IR) of PH required calculation of years of follow-up for each patient. For details refer to on-line supplement S1. Incidence rate ratios were compared to assess differences in IR between the three age groups. To evaluate ALT, GGT, and ALP as possible early risk predictors it was the intention to combine the data from the 2 sites. Details are provided in Table S2. To estimate the effects of 2 or more elevated ALT, GGT, and ALP values (recorded as 0 (not elevated) and 1 (elevated) separately on the risk of developing PH, Cox proportional hazard models were conducted for the unadjusted enzyme data and for the data set adjusted for number of tests and stratified by clinic [24]. Similar tests were conducted for the presence of one abnormal enzyme and 2 or more abnormal enzymes for the combinations ALT/ALP; ALT/GGT and ALP/GGT. Sample size for multivariate analysis was based on the recommendation that the population of PH should be at least 10× the number of covariates (the 3 enzymes) i.e. 30 cases [25]. With 51 cases overall and 47 in the enzyme cohort the study has sufficient power to provide an interpretable outcome. SPSS V25 was used for all analyses and alpha was set to 0.05. The study was approved by the Human Research Ethics Committees at both centres. Results Overview of combined and individual clinic data for total and enzyme subsets From January 1st 1986 to January 1st 2007, 577 infants were diagnosed with CF at newborn screening, (278 VCFC and 299 CHW) (Table 1). A total of 51/577 (8.8%) developed PH (24/278 (8.6%) VCFC, 27/299(9.0%) CHW) (p = 0.87) inclusive of those until 18.5 years of age or to January 2014 (the nominated end of the data collection), or to age of PH or death/transplant (MTx). Similar proportions of PH were found for the enzyme subset (p = 0.86). There were no significant differences in gender between those with and without PH (total cohort p = 0.06, enzyme subset p = 0.37) although there was a trend to male predominance in the PH subgroup of the total cohort. There were also no differences in the number of tests between those with and without PH (total cohort p = 0.24, enzyme subset p = 0.44). Differences between the two sites on demographic and clinical variables are reported in Table S2. Children at VCFC had more tests when ≤6.5 years of age (total cohort and enzyme subset p < 0.001) and VCFC had a higher proportion with 2 or more elevated GGT (p = 0.04). This higher proportion with abnormal GGT at VCFC is possibly related to the increased number of tests conducted at this clinic. All PH patients except one were PI and had 2 severe type 1 to type 111mutations. The exception with PS had a DF508/unknown genotype.
Please cite this article as: C. Marco, F. Judith and W. Donna et al., Occurrence, outcomes and predictors of portal hypertension in cystic fibrosis: A longitudinal prospective birth cohort study, Journal of Cystic Fibrosis, https://doi.org/10.1016/j.jcf.2019.09.016
ARTICLE IN PRESS
JID: JCF
[m5G;October 29, 2019;19:29]
C. Marco, F. Judith and W. Donna et al. / Journal of Cystic Fibrosis xxx (xxxx) xxx
3
Table 1 Total cohort and enzyme subset. Total cohort N = 577
Characteristic
Enzyme subset N = 512
All N = 577
With PH N = 51
Without PH N = 526
p value
All N = 512
With PH N = 47
Without PH N = 465
p value
Site – no. (%) Italy (VCFC) Australia (CHW)
278 (48.2) 299 (51.8)
24/278 (8.6) 27/299 (9.0)
254 (48.3) 272 (51.7)
0.87a
257 (50.2) 255 (49.8)
23/257 (8.9) 24/255 (9.4)
234 (50.3) 231 (49.7
0.86a
Gender – no. (%) Male Female
268 (46.4) 309 (53.6)
30 (58.8) 21 (41.2)
238 (45.2) 288 (54.8)
0.06a
251 (49.0) 261 (51.0)
26 (55.3) 21 (44.7)
255 (51.5) 240 (48.5)
0.37a
No. tests when ≤ 6.5 years- Mean years ±SD Both locations
5.0 ± 2.2
4.6 ± 2.2
0.24b
5.4 ± 2.0
5.2 ± 1.8
0.44b
Age to PH/end of study Mean years ±SD Both locations
10.65 ± 4.05
14.6 ± 4.5
<0.001b
10.9 ± 3.9
14.6 ± 4.3
a b
Chi Square Test of Independence. Independent Group t test.
Table 2 Comparison of incidence rates for development of PH by age group, combined clinics, total cohort. Age group
1 2 a b c d
<0.001b
1 3 to <8 years n = 11 IR (95% CI) 2.81a (1.40–5.03)
2 8–<13 years n = 23 IR (95% CI) 3.95a (2.56–5.83)
3 13–<18 years n = 17 IR (95% CI) 2.55a (1.52–3.95)
– –
0.71b (0.33–1.52) –
1.10c (0.49–2.45) 1.55d (0.82–2.91)
Incidence rates per 1,0 0 0 patient years of followup and associated 95% CI. Incidence rate ratio IRR (95% CI) 3–<8 yrs/8–<13 years. Incidence rate ratio IRR (95% CI) 3–<8 yrs/13–<18 years. Incidence rate ratio IRR (95% CI) 8–<13 yrs/13 <18 years.
Prevalence and incidence rates of PH The mean age of onset of PH was 10.65 SD 4.05 years (Table 1) but between the clinics there was a significantly younger age of onset of 9.34 SD 4.09 years at CHW versus 12.13 SD 3.52 years VCFC (p = 0.01) (Table S2). The incidence rate of PH was calculated in 3 age groups (3–7; 8–12 and 13–18 years) for the total cohort where the average rate peaked in mid-childhood near 4/10 0 0 patient years (Table 2). Separate clinic data are reported in Table S3. Incidence rates were not significantly different between age groups in the total cohort or between the separate clinic data.
MTx occurrence The other major outcome assessed on the total cohort was combined mortality and transplant. Although overall MTx was low in the total cohort (37/577, 6.4%) there was a significantly higher occurrence in PH 12/51 (23.5%) versus non-PH 25/526 (4.8%) (p < 0.001). In the PH clinic subgroups MTx occurred in 3/24 (12.5%) VCFC and 9/27 (33.3%) CHW but the difference was not significant (Fishers exact p = 0.11). At VCFC there were 2 liver transplants and 1 lung transplant and 0 deaths; at CHW 5 liver transplants and 4 deaths–2 end stage respiratory disease and 2 with lung and liver failure. Of the 51 PH patients, those with MTx (n = 12) had a significantly younger mean age of onset of PH, 7.97 SD 2.80 years compared to non-MTx PH patients (n = 39) 11.47 SD 4.04 years (p = 0.007). Oesophageal variceal bleeding occurred in 7 patients requiring variceal banding and another patient developed hepatic encephalopathy [5].
Table 3 Mantel Haenszel tests for association between two or more elevated LFTs and portal hypertension, enzyme subgroup. Number with PH / Number of patients (%) ALT ≥2 abnormal tests Yes No p value ALP ≥2 abnormal tests Yes No p value GGT ≥2 abnormal tests Yes No p value a
16/9931/413
(16.2)(7.5)
0.01a 16/12031/392
(13.3)(7.9)
0.10a 20/7527/437
(26.7)( 6.2)
<0.001a
Mantel-Haenszel, stratified by clinic.
Enzyme data A summary of the association between elevated LFTs (in the first 6.5 years of life) and PH after stratifying by clinic and adjusting for the number of tests is shown in Table 3. Of the 99 patients with 2 or more abnormal ALT tests 16 (16.2%) had PH while of the 413 without elevated ALTs 31 (7.5%) had PH (p = 0.01). The association between 2 or more elevated GGTs and PH demonstrated 20/75 (26.7%) developed PH while 27/437 (6.2%) with normal GGT developed PH (p < 0.001). The association between 2 or more abnormal ALPs and PH was not significantly different than those without 2 abnormal ALPs (p = 0.10). Over half of those with PH had no evidence of enzyme elevation on 2 or more occasions and 5/47 (10.6%) with PH never demonstrated any enzyme abnormalities. Hazard ratios for 2 or more elevated results and the development of PH unadjusted and adjusted for clinic and number of tests are reported in Table 4. Patients with elevated ALT, ALP or GGT on 2 or more occasions were 1.86 to 5.71 times more likely to develop PH than those with less than 2 elevated results. Interpretation did not differ between adjusted and non-adjusted results for ALT and GGT while results for ALP were significant only in the adjusted/stratified analysis. To further examine these findings, hazard ratios were also calculated for (a) one abnormal result for ALT, ALP and GGT individually; and (b) 2 or more abnormal results for any 2 test combina-
Please cite this article as: C. Marco, F. Judith and W. Donna et al., Occurrence, outcomes and predictors of portal hypertension in cystic fibrosis: A longitudinal prospective birth cohort study, Journal of Cystic Fibrosis, https://doi.org/10.1016/j.jcf.2019.09.016
ARTICLE IN PRESS
JID: JCF 4
[m5G;October 29, 2019;19:29]
C. Marco, F. Judith and W. Donna et al. / Journal of Cystic Fibrosis xxx (xxxx) xxx Table 4 Cox proprortional hazard hazard ratios for two or more elevated LFTs and development of portal hypertension, enzyme subgroup. ≥2 Elevated liver function tests
Unadjusted hazard ratios (95% CI)
Adjusted Hazard ratios (95% CI)∗
ALT Yes No p value
2.44 (1.33–4.47)1.00
2.69 (1.43–5.04)1.00
0.004
0.002
ALP Yes No p value
1.73 (0.95–3.17)1.00
1.86 (1.00–3.46)1.00
0.074
0.049
GGT Yes No p value
2.80 (1.19–6.86)1.00
5.71 (3.11–10.47)1.00
0.019
<0.001
∗
Stratified by clinic and adjusted for number of tests.
tions (ALT/ALP, ALT/GGT, GGT/ALP). One abnormal test only was not associated with PH for ALT (n = 87) (HR 1.55; 95% CI 0.79 to 3.03, p = 0.24) or ALP (n = 146) (HR 1.51; 95% CI: 0.83 to 2.74, p = 0.18) but was negatively associated with GGT (n = 304) (HR 0.37; 95% CI 0.20 to 0.67; p = 0.001). There was a higher risk of developing PH with combinations than with individual enzymes: ALT/ALP (n = 44) (HR 3.59; 95% CI, 1.76 to 7.31; p < 0.001); ALT/GGT (n = 39) (HR 5.56; 95% CI 2.82 to 10.98: p < 0.001) and ALP/GGT (n = 30) (HR 5.74 95% CI 2.78 to 11.86, p < 0.001) (see Table S4 for separate clinic results). Kaplan-Meier survival curves and the log rank test results for the 3 liver enzymes individually are depicted in Figure S1a – S1c and show the earlier onset of PH for those with 2 or more elevated enzymes on at least 2 occasions. Discussion This study was conducted on a large cohort of CF infants diagnosed by NBS and referred to one of the two participating CF centres which have attending gastroenterologists and provides accurate and reliable data on the occurrence and onset of PH as evident by the appearance of splenomegaly and confirmed by imaging with ultrasound or NMR. The study represents the first time a CF neonatal cohort has been followed to 18.5 years, assessing the occurrence and subsequent evolution of severe liver disease. The results highlight an overall prevalence of 8.8% in the combined clinic data (8.6% VCFC and 9.0% CHW) with incident rates of 2.81/10 0 0 patient years (3–7 years of age), 3.95/10 0 0 patient years (8–12 years of age) and 2.55/10 0 0 patient years (13–18 years of age) and a significantly higher (p < 0.001) mortality/transplant occurrence in those with severe liver disease with PH as compared with the non-PH subgroup. Multiple studies have assessed prevalence of cirrhosis with PH and a wide range has been found from 1% to 22% but these extreme values are likely underestimations or overestimations (respectively) and most of the larger studies have a prevalence in the 5–7% range. In this regard the most recent and one of the largest studies of CFPI patients is the French Gene Modifier Study (FGMS) [26] of severe liver disease (SLD) which reported a prevalence of 175/3328 (5.26%), 60% of that found in the current study. When comparing these two studies the French study used a less conservative definition of SLD that included a majority of patients (56%) without PH, and PH occurred in only 44% of the total 175 with SLD and therefore prevalence of PH was only 2.3% compared with 8.8% in the present study. The incidence rates would also be decreased accordingly. This marked difference could be explained by the fact that annual examination in France only started in 2001, 16 years after annual data collection commenced. The latter questions the
reliability of the data prior to 2001 and could contribute to the rising incidence in the 20–30 year old group after mandatory annual examination commenced, i.e. there was a delay in diagnosis until transition to adult centres or to the adult section of a CF centre. Obviously this would not account for the difference between FGMS and our study unless in the FGMS some patients developed severe lung disease and died during childhood, adolescence or early adulthood either prior to the development of PH or with unrecognized PH to account for the difference between their study and ours, but data on mortality and transplants were not provided to support or dispute this possibility. Combined mortality/transplant (MTx) data in the present study was over 4x higher (p < 0.001) in the PH versus the non-PH subgroup in accord with the match controlled Irish cohort study [27]. Of the 51 PH patients, those with MTx had a significantly lower age of onset of PH of 8 years compared to 12 years for those without MTx (p = 0.007) supporting the suggestion of other studies [2,11,13,15] that severe morbidity or mortality and aggressive liver disease occurs in patients with early onset PH i.e. it is a midchildhood phenomenon and de novo occurrence in adult patients is rare or non-existent [11,13,15,29]. The presence of PH is thus required to justify inclusion in the SLD category as PH is the marker of SLD used in several studies [2,3,13,27–29] including the International Gene Modifier study [2]. It is possible that PH and the requirement for liver transplant following liver failure may increase over 30 years of age. The above discussion has not accounted for the occurrence of Nodular Regenerative Hyperplasia (NRH) a recently described entity over the last 12 months [30]. They describe PH occurring in over 25 patients with CF and PH without liver cirrhosis on ex-plants but with PV occlusion of varying severity. We have not seen this entity on our ex-plant pathology but all CF centres should be aware of this phenomena to investigate its origin. The individual liver enzymes ALT and GGT and the combinations ALT/GGT, GGT/ALP and ALT/ALP if elevated on two or more occasions in early life were significantly predictive of the later development of PH. Whether the clinics were combined or analysed separately and regardless of the greater number of tests done in Verona, in general a similar pattern emerged in relation to abnormal LFTs and PH. As less than 30% of PH patients exhibited these patterns they were not useful for predicting PH in individual patients. However, the elevated enzyme patterns in earlier life were significantly associated with earlier onset of PH. These findings suggest that there are two emerging phenotypes in CFSLD: firstly a severe phenotype with early onset MBC with PH with complications including variceal bleeding and in some early liver failure necessitating liver transplantation, and secondly a milder phenotype with MBC and PH who have a more benign course with minimal
Please cite this article as: C. Marco, F. Judith and W. Donna et al., Occurrence, outcomes and predictors of portal hypertension in cystic fibrosis: A longitudinal prospective birth cohort study, Journal of Cystic Fibrosis, https://doi.org/10.1016/j.jcf.2019.09.016
JID: JCF
ARTICLE IN PRESS
[m5G;October 29, 2019;19:29]
C. Marco, F. Judith and W. Donna et al. / Journal of Cystic Fibrosis xxx (xxxx) xxx
LFT disturbance, less complications and minimal or no requirement for transplantation extending into adulthood until at least 30 years of age. Although this study was based on a total cohort collected in two clinics over 20 years, the small sample size in subgroups, such as gender and age group, has potentially resulted in a lack of statistical significance. Given median survival is now extending into the late 40’s a large multicentred international study needs to be performed to address the issues outlined above. Declaration of Competing Interest The authors declare that there are no conflicts of interest. Acknowledgements The authors would like to acknowledge Ms Cheryl Frazer and Mrs Margie Gruca for their tireless efforts in retrieving data, maintaining the database and formatting the manuscript for publication. Supplementary materials Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.jcf.2019.09.016. References [1] Sokol RJ, Durie PR. Recommendations for management of liver and biliary tract disease in cystic fibrosis. Cystic Fibrosis Foundation Hepatobiliary Disease Consensus Group. J Pediatr Gastroenterol Nutr 1999;28:S1–3. [2] Stonebraker JR, Ooi CY, Pace RG, Corvol H, Knowles MR, Durie PR, Ling SC. Features of severe liver disease with portal hypertension in patients with cystic fibrosis. Clin Gastroenterol Hepatol 2016;14:1207–15. [3] Flass T, Narkewicz MR. Cirrhosis and other liver disease in cystic fibrosis. J Cyst Fibrosis 2013;12:116–24. [4] Ling SC, Wilkinson JD, Hollman AS, McColl J, Evans TJ, Paton JY. The evolution of liver disease in cystic fibrosis. Arch Dis Child 1999;81:129–32. [5] Pandit C, Graham C, Selvadurai H, Gaskin K, Cooper P, Van Asperen P. Festival food coma in cystic fibrosis. Pediatric Pulmonol 2013;48:725–7. [6] Nightingale S, O’Loughlin EV, Dorney SF, et al. Isolated liver transplantation in children with cystic fibrosis – an Australian experience. Pediatr Transplant 2010;14:779–85. [7] Leung DH, Ye W, Molleston JP, et al. Baseline ultrasound and clinical correlates in children with cystic fibrosis. J Pediatr 2015;167:862–8. [8] Lenaerts C, Lapierre C, Patriquin H, et al. Surveillance for cystic fibrosis-associated hepatobiliary disease: early ultrasound changes and predisposing factors. J Pediatr 2003;143:343–50.
5
[9] Lamireau T, Monnereau S, Martin S, Marcotte J, Winnock M, Alvarez F. Epidemiology of liver disease in cystic fibrosis: a longitudinal study. J Hepatol 2004;41:920–5. [10] Fitzsimmons. Cystic fibrosis foundation patient registry. Elsevier; 1997. Annual data base report Bethesda, Maryland. [11] Scott-Jupp R, Lama M, Tanner MS. Prevalence of liver disease in cystic fibrosis. Arch Dis Child 1991;66:698–701. [12] Vawter GF, Shwachman H. Cystic fibrosis in adults, an autopsy study. Pathol Annual 1979;14:357–82. [13] Nash KL, Allison ME, McKeon D, et al. A single centre experience of liver disease in adults with cystic fibrosis 1995–2006. J Cyst Fibrosis 2008;7:252–7. [14] Feigelson J, Anagnostopoulos C, Poquet M, Pecau Y, Munck A, Navarro J. Liver cirrhosis in cystic fibrosis – therapeutic implications and long term follow up. Arch Dis Child 1993;68:653–7. [15] Colombo C, Battezzati PM, Crosignani A, Morabito A, Costantini D, Padoan R. Liver disease in cystic fibrosis: a prospective study on incidence, risk factors and outcome. Hepatology 2002;36:1374–82. [16] Corbett K, Kelleher S, Rowland M, et al. Cystic fibrosis-associated liver disease: a population-based study. J Pediatr 2004;145:327–32. [17] Mayer-Hamblett N, Kloster M, Ramsey BW, Narkewicz MR, Saiman L, Goss CH. Incidence and clinical significance of elevated liver function tests in cystic fibrosis clinical trials. Contemp Clin Trials 2013;34:232–8. [18] Wilschanski M, Rivlin J, Cohen S, et al. Clinical and genetic risk factors for cystic fibrosis-related liver disease. Pediatrics 1999;103:52–7. [19] Narkewicz MR. Markers of cystic fibrosis-associated liver disease. J Pediatr Gastroenterol Nutr 2001;32:421–2. [20] Cipolli M, Castellani C, Wilcken B, Massie J, McKay K, Gruca M, et al. Pancreatic phenotype in infants with cystic fibrosis identified by two neonatal mutational screening programs. Arch Dis Child 2007;92:842–6. [21] Waters DL, Dorney SFA, Gaskin KJ, Gruca MA, O’Halloran M, Wilcken B. Pancreatic function in infants identified as having cystic fibrosis in a neonatal screening program. New Engl J Med 1990;322:303–8. [22] Soldin SJ. Pediatric reference ranges. Soldin SJ, editor. second ed. Washington DC: AACC Press; 1997. [23] George J, Denney-Wilson E, Okely AD, Hardy LL, Aitken R. The population distributions, upper normal limits and correlations between liver tests among Australian adolescents. J Paediatr Child Health 2008;44:579–85. [24] Altman DG, Bland JM. Interaction revisited: the difference between two estimates. BMJ 2003;326(7382):219. [25] Norman GE, Streiner DL. Biostatistics: the bare essentials. St Louis: Mosby Year Book Inc.; 1994. p. 108–18. [26] Boelle P.-Y., Debray D., Guillot L., Clement A., Corvol H. Cystic fibrosis liver disease: outcomes and risk factors in a large cohort of French patients. doi:10. 1002/hep.30148. [27] Rowland M, Gallagher CG, O’Laoide R, et al. Outcome in cystic fibrosis liver disease. Am J Gastroenterol 2011;106:104–9. [28] Melzi ML, Kelly DA, Colombo C, Jara P, Manzanares J, Colledan M, et al. Liver transplant in cystic fibrosis: a poll among European centers. A study from the European Liver Transplant Registry. Transpl Int 2006;19:726–31. [29] Desmond CP, Wilson J, Bailey M, Clark D, Roberts SK. The benign course of liver disease in adults with cystic fibrosis and the effect of ursodeoxycholic acid. Liver Int 2007;27:1402–8. [30] Wu H, Vu M, Dhingra S, Ackah R, Goss JA, Rana A, Quintanilla N, Patel K, Leung DH. Obstructive portal venopathy without cirrhosis is prevalent in pediatric cystic fibrosis liver disease with portal hypertension. Clin Gastroenterol Hepatol 2019;17:2134–6.
Please cite this article as: C. Marco, F. Judith and W. Donna et al., Occurrence, outcomes and predictors of portal hypertension in cystic fibrosis: A longitudinal prospective birth cohort study, Journal of Cystic Fibrosis, https://doi.org/10.1016/j.jcf.2019.09.016
JID: JCF
[m5G;October 29, 2019;19:29]
Journal of Cystic Fibrosis xxx (xxxx) xxx
Contents lists available at ScienceDirect
Journal of Cystic Fibrosis journal homepage: www.elsevier.com/locate/jcf
Occurrence, outcomes and predictors of portal hypertension in cystic fibrosis: A longitudinal prospective birth cohort study Cipolli Marco a, Fethney Judith b, Waters Donna b, Zanolla Luisa c, Meneghelli Ilaria a, Shoma Dutt d, Assael Baroukh Maurice e, Gaskin Kevin John d,∗ a
Cystic Fibrosis Center, Azienda Ospedaliera Universitaria Integrata, Verona, Italy Susan Wakil School of Nursing and Midwifery, University of Sydney, Sydney, Australia Cardiology Department, Azienda Ospedaliera Universitaria Integrata, Verona, Italy d Department of Gastroenterology/James Fairfax Institute of Paediatric Nutrition, The Children’s Hospital at Westmead, Sydney, Australia e Scientific Advisor, Adult Cystic Fibrosis Center, University of Milano Medical School, Milano, Italy b c
a r t i c l e
i n f o
Article history: Received 4 April 2019 Revised 29 August 2019 Accepted 24 September 2019 Available online xxx Keywords: Portal hypertension Liver enzymes Incidence/Prevalence
a b s t r a c t Background: The reported prevalence of portal hypertension (PH) in Cystic Fibrosis is variable, incidence rates rarely provided and the utility of liver function tests (LFT’s) early in life to predict PH is questionable. The aims were to (1) determine PH prevalence (P) and incidence rate (IR) and combined mortality transplant (MTX) data in PH vs non-PH patients and (2) to assess association of LFTs in early life with liver disease and PH. Method: (1) A double centre longitudinal cohort study of 577 CF patients diagnosed by newborn screening (NBS) with annual examinations for PH up to 18.5 years of age (max) was performed over 28 years for P, IR, and MTX data; (2) Cox proportional hazard models were used to assess the association of elevated LFTs on 2 or more occasions over 0–6.5 years and PH. Results: 51/577(8.8%) developed PH with an average IR of near 3/10 0 0 patient years per 5 year interval representing young, mid and late childhood respectively in patients 3–18 years of age. Combined mortality/liver transplant occurred in 12/51 (23.5%) PH and 25/526 (4.8%) non-PH (p < 0.001). Elevated enzymes particularly GGT (HR:5.71, 95% CI 3.11–10.47); ALT/GGT (HR: 5.56, 95% CI 2.82–10.98); and ALP/GGT (HR: 5.74, 95% CI 2.78–11.86) were associated with the onset of PH. Conclusion: This birth cohort with annual examination for PH provides an accurate assessment of the prevalence, and IR of PH and MTX of PH vs non-PH. Early elevated LFTs are associated with onset of MBC/PH. © 2019 European Cystic Fibrosis Society. Published by Elsevier B.V. All rights reserved.
Introduction The severe form of CF liver disease, multilobular biliary cirrhosis (MBC) is associated with the almost invariable development of portal hypertension (PH) and the complications of variceal bleeding [1–4] and unusually hepatic encephalopathy [5] or liver fail-
Abbreviations: ALP, alkaline phosphatase; ALT, alanine aminotransferase; CF, cystic fibrosis; GGT, gamma glutamyl transpeptidase; HR, hazard ratio; LFTs, liver function tests; MBC, multilobular biliary cirrhosis; MTx, mortality/transplant; PH, portal hypertension. ∗ Corresponding author at: The Children’s Hospital at Westmead, Locked Bag 4001, Westmead, NSW 2145, Australia. E-mail addresses: [email protected] (C. Marco), [email protected] (F. Judith), [email protected] (W. Donna), [email protected] (Z. Luisa), [email protected] (M. Ilaria), [email protected] (S. Dutt), [email protected] (A. Baroukh Maurice), [email protected] (G. Kevin John).
ure requiring transplantation [6]. These features in combination with splenomegaly and laboratory evidence of hypersplenism facilitate a diagnosis of MBC/PH which can be confirmed with noninvasive ultrasonography [2,7–9] and magnetic resonance imaging. Although the average prevalence in large studies is 5–6% [1,3,9– 16] extreme values of 1% [10,11] to 22% [12,13] have been reported. The latter could represent a marked heterogeneity of MBC/PH but they are likely related to the different study designs/methodology or the age groups being studied. The 1% value was from a postal questionnaire [10] and was subsequently considered an underestimation related to the accuracy of data recorded and collected [1]. Regarding the extreme 2 values near 20%, one was an autopsy study [12] and the other was from a transplant centre [13], with the prevalence calculated from the smaller group undergoing autopsy or being considered for transplant rather than the true population of origin. A more reliable estimate could be obtained from
https://doi.org/10.1016/j.jcf.2019.09.016 1569-1993/© 2019 European Cystic Fibrosis Society. Published by Elsevier B.V. All rights reserved.
Please cite this article as: C. Marco, F. Judith and W. Donna et al., Occurrence, outcomes and predictors of portal hypertension in cystic fibrosis: A longitudinal prospective birth cohort study, Journal of Cystic Fibrosis, https://doi.org/10.1016/j.jcf.2019.09.016
JID: JCF 2
ARTICLE IN PRESS
[m5G;October 29, 2019;19:29]
C. Marco, F. Judith and W. Donna et al. / Journal of Cystic Fibrosis xxx (xxxx) xxx
clinics utilising newborn screening for CF diagnosis and the birth cohort number as the true population of origin. Several publications [1,3,17–19] have reported on the futility of using the common serum liver enzyme tests (LFTs) alanine aminotransferase (ALT), gamma glutamyl transpeptidase (GGT) and alkaline phosphatase (ALP) to predict the occurrence of PH in CF. However, another report [15] suggested that if any 2 enzymes were elevated on 2 or more occasions in the first 5 years of life this scenario was over 6x more common in those with PH than in those without PH in later childhood, but no statistical evaluation was provided. LFTs therefore require further evaluation to assess their utility in predicting the later occurrence/outcomes of PH and liver disease. A collaborative database was developed from 2 of the original longterm neonatal screening clinics, Verona CF centre (VCFC), Italy and the Children’s Hospital at Westmead (CHW) CF Clinic, Australia and contains annual data from birth on both examination signs of MBC/PH and LFTs from 1986–2014.The aims of this study were first to provide reliable prevalence and incidence rates of PH and outcomes (mortality/ liver transplant rates) in comparison with the non-PH sub-group and secondly to determine whether elevated serum LFTs on 2 or more occasions in the first 6.5 years of life are associated with the development of MBC/PH and complications. Methods Study population A longitudinal cohort study was conducted on all CF patients diagnosed by newborn screening (NBS) and referred to VCFC and CHW from January 1st 1986 to January 1st 2007. Patients were followed to a maximum 18.5 years of age or January 2014, date of death or liver transplant whichever was sooner. Genotypically, the centres differed (VCFC < 30% DF508 homozygotes, 50% DF508/other; CHW > 50% DF508 homozygotes, 43% DF508/other) noting the proportions with severe type 1-111 mutations and the mild type1V and V mutations were very similar as were the proportions with pancreatic insufficiency (PI) sufficiency (PS) and neonatal meconium ileus at both centres [20–21]. All patients had confirmatory sweat chlorides with compound heterozygotes having extended mutation analyses as available in each centre. MBC/PH was defined clinically at annual examinations as the presence of a palpable firm multilobulated liver (>90% of cases) and splenomegaly (100%). All PH cases were confirmed initially by ultrasound regarding the presence of a multilobulated liver and splenomegaly and the variable presence of intraabdominal varices and spontaneous porto-systemic shunts. No cases had portal vein thrombosis nor portal cavernomas. MRI angiography was undertaken in less than 10% where the portal vein could not be adequately visualised on ultrasound examination. Endoscopic examination for oesophageal/gastric varices was used rarely to confirm imaging findings but was used in every case with haematemesis (3 cases occurring within the first 3 years after the onset of PH in the 8–12 age range and in 4 cases occurring in 13–16 years of age whose PH was diagnosed 11–12 years of age). Other causes of cirrhosis such as infectious and autoimmune hepatitis, alpha1-antitrypsin deficiency and Wilsons disease were excluded. MTx was the combined mortality and transplants at both centres as numbers in the PH subgroup were low (7 liver Tx, 1 lung Tx and 4 deaths). To evaluate the 3 liver enzymes (ALT, GGT and ALP) as potential early predictors of MBC/PH, the LFTs were collected at diagnosis and annually, with enzyme data censored at 6.0 years of age (allowing a max of 6.5 years for delays in annual exams) or until the time of death or onset of PH if occurring pre-6.5 years. This timepoint was chosen due to (a) previous reporting, but without
statistical analysis, of LFT elevation 0–5 years of life and the later occurrence of CFLD [15] and (b) 6.0 years of age being approximately half the average age of the peak occurrence of MBC [29]. During this interval there were 31/577 (5.4%) patients with no LFT data. Of the remaining 546 patients, 512 had LFTs measured on 2 or more of their annual visits. Reference values for normal children were derived from pooled samples at varying ages with cross referencing to the paediatric literature and elevated values >4 weeks of age defined as above the ULN: ALT 50 IU/L, GGT 25 IU/L and ALP 350 IU/L [22–23]. Statistical analysis Prevalence of PH was (1) the proportion of patients who had developed PH by 1st January 2014 for the total population of 577 patients; (2) the enzyme cohort of 512 patients; and (3) the corresponding data for the 2 clinics separately. Incidence rates (IR) of PH required calculation of years of follow-up for each patient. For details refer to on-line supplement S1. Incidence rate ratios were compared to assess differences in IR between the three age groups. To evaluate ALT, GGT, and ALP as possible early risk predictors it was the intention to combine the data from the 2 sites. Details are provided in Table S2. To estimate the effects of 2 or more elevated ALT, GGT, and ALP values (recorded as 0 (not elevated) and 1 (elevated) separately on the risk of developing PH, Cox proportional hazard models were conducted for the unadjusted enzyme data and for the data set adjusted for number of tests and stratified by clinic [24]. Similar tests were conducted for the presence of one abnormal enzyme and 2 or more abnormal enzymes for the combinations ALT/ALP; ALT/GGT and ALP/GGT. Sample size for multivariate analysis was based on the recommendation that the population of PH should be at least 10× the number of covariates (the 3 enzymes) i.e. 30 cases [25]. With 51 cases overall and 47 in the enzyme cohort the study has sufficient power to provide an interpretable outcome. SPSS V25 was used for all analyses and alpha was set to 0.05. The study was approved by the Human Research Ethics Committees at both centres. Results Overview of combined and individual clinic data for total and enzyme subsets From January 1st 1986 to January 1st 2007, 577 infants were diagnosed with CF at newborn screening, (278 VCFC and 299 CHW) (Table 1). A total of 51/577 (8.8%) developed PH (24/278 (8.6%) VCFC, 27/299(9.0%) CHW) (p = 0.87) inclusive of those until 18.5 years of age or to January 2014 (the nominated end of the data collection), or to age of PH or death/transplant (MTx). Similar proportions of PH were found for the enzyme subset (p = 0.86). There were no significant differences in gender between those with and without PH (total cohort p = 0.06, enzyme subset p = 0.37) although there was a trend to male predominance in the PH subgroup of the total cohort. There were also no differences in the number of tests between those with and without PH (total cohort p = 0.24, enzyme subset p = 0.44). Differences between the two sites on demographic and clinical variables are reported in Table S2. Children at VCFC had more tests when ≤6.5 years of age (total cohort and enzyme subset p < 0.001) and VCFC had a higher proportion with 2 or more elevated GGT (p = 0.04). This higher proportion with abnormal GGT at VCFC is possibly related to the increased number of tests conducted at this clinic. All PH patients except one were PI and had 2 severe type 1 to type 111mutations. The exception with PS had a DF508/unknown genotype.
Please cite this article as: C. Marco, F. Judith and W. Donna et al., Occurrence, outcomes and predictors of portal hypertension in cystic fibrosis: A longitudinal prospective birth cohort study, Journal of Cystic Fibrosis, https://doi.org/10.1016/j.jcf.2019.09.016
ARTICLE IN PRESS
JID: JCF
[m5G;October 29, 2019;19:29]
C. Marco, F. Judith and W. Donna et al. / Journal of Cystic Fibrosis xxx (xxxx) xxx
3
Table 1 Total cohort and enzyme subset. Total cohort N = 577
Characteristic
Enzyme subset N = 512
All N = 577
With PH N = 51
Without PH N = 526
p value
All N = 512
With PH N = 47
Without PH N = 465
p value
Site – no. (%) Italy (VCFC) Australia (CHW)
278 (48.2) 299 (51.8)
24/278 (8.6) 27/299 (9.0)
254 (48.3) 272 (51.7)
0.87a
257 (50.2) 255 (49.8)
23/257 (8.9) 24/255 (9.4)
234 (50.3) 231 (49.7
0.86a
Gender – no. (%) Male Female
268 (46.4) 309 (53.6)
30 (58.8) 21 (41.2)
238 (45.2) 288 (54.8)
0.06a
251 (49.0) 261 (51.0)
26 (55.3) 21 (44.7)
255 (51.5) 240 (48.5)
0.37a
No. tests when ≤ 6.5 years- Mean years ±SD Both locations
5.0 ± 2.2
4.6 ± 2.2
0.24b
5.4 ± 2.0
5.2 ± 1.8
0.44b
Age to PH/end of study Mean years ±SD Both locations
10.65 ± 4.05
14.6 ± 4.5
<0.001b
10.9 ± 3.9
14.6 ± 4.3
a b
Chi Square Test of Independence. Independent Group t test.
Table 2 Comparison of incidence rates for development of PH by age group, combined clinics, total cohort. Age group
1 2 a b c d
<0.001b
1 3 to <8 years n = 11 IR (95% CI) 2.81a (1.40–5.03)
2 8–<13 years n = 23 IR (95% CI) 3.95a (2.56–5.83)
3 13–<18 years n = 17 IR (95% CI) 2.55a (1.52–3.95)
– –
0.71b (0.33–1.52) –
1.10c (0.49–2.45) 1.55d (0.82–2.91)
Incidence rates per 1,0 0 0 patient years of followup and associated 95% CI. Incidence rate ratio IRR (95% CI) 3–<8 yrs/8–<13 years. Incidence rate ratio IRR (95% CI) 3–<8 yrs/13–<18 years. Incidence rate ratio IRR (95% CI) 8–<13 yrs/13 <18 years.
Prevalence and incidence rates of PH The mean age of onset of PH was 10.65 SD 4.05 years (Table 1) but between the clinics there was a significantly younger age of onset of 9.34 SD 4.09 years at CHW versus 12.13 SD 3.52 years VCFC (p = 0.01) (Table S2). The incidence rate of PH was calculated in 3 age groups (3–7; 8–12 and 13–18 years) for the total cohort where the average rate peaked in mid-childhood near 4/10 0 0 patient years (Table 2). Separate clinic data are reported in Table S3. Incidence rates were not significantly different between age groups in the total cohort or between the separate clinic data.
MTx occurrence The other major outcome assessed on the total cohort was combined mortality and transplant. Although overall MTx was low in the total cohort (37/577, 6.4%) there was a significantly higher occurrence in PH 12/51 (23.5%) versus non-PH 25/526 (4.8%) (p < 0.001). In the PH clinic subgroups MTx occurred in 3/24 (12.5%) VCFC and 9/27 (33.3%) CHW but the difference was not significant (Fishers exact p = 0.11). At VCFC there were 2 liver transplants and 1 lung transplant and 0 deaths; at CHW 5 liver transplants and 4 deaths–2 end stage respiratory disease and 2 with lung and liver failure. Of the 51 PH patients, those with MTx (n = 12) had a significantly younger mean age of onset of PH, 7.97 SD 2.80 years compared to non-MTx PH patients (n = 39) 11.47 SD 4.04 years (p = 0.007). Oesophageal variceal bleeding occurred in 7 patients requiring variceal banding and another patient developed hepatic encephalopathy [5].
Table 3 Mantel Haenszel tests for association between two or more elevated LFTs and portal hypertension, enzyme subgroup. Number with PH / Number of patients (%) ALT ≥2 abnormal tests Yes No p value ALP ≥2 abnormal tests Yes No p value GGT ≥2 abnormal tests Yes No p value a
16/9931/413
(16.2)(7.5)
0.01a 16/12031/392
(13.3)(7.9)
0.10a 20/7527/437
(26.7)( 6.2)
<0.001a
Mantel-Haenszel, stratified by clinic.
Enzyme data A summary of the association between elevated LFTs (in the first 6.5 years of life) and PH after stratifying by clinic and adjusting for the number of tests is shown in Table 3. Of the 99 patients with 2 or more abnormal ALT tests 16 (16.2%) had PH while of the 413 without elevated ALTs 31 (7.5%) had PH (p = 0.01). The association between 2 or more elevated GGTs and PH demonstrated 20/75 (26.7%) developed PH while 27/437 (6.2%) with normal GGT developed PH (p < 0.001). The association between 2 or more abnormal ALPs and PH was not significantly different than those without 2 abnormal ALPs (p = 0.10). Over half of those with PH had no evidence of enzyme elevation on 2 or more occasions and 5/47 (10.6%) with PH never demonstrated any enzyme abnormalities. Hazard ratios for 2 or more elevated results and the development of PH unadjusted and adjusted for clinic and number of tests are reported in Table 4. Patients with elevated ALT, ALP or GGT on 2 or more occasions were 1.86 to 5.71 times more likely to develop PH than those with less than 2 elevated results. Interpretation did not differ between adjusted and non-adjusted results for ALT and GGT while results for ALP were significant only in the adjusted/stratified analysis. To further examine these findings, hazard ratios were also calculated for (a) one abnormal result for ALT, ALP and GGT individually; and (b) 2 or more abnormal results for any 2 test combina-
Please cite this article as: C. Marco, F. Judith and W. Donna et al., Occurrence, outcomes and predictors of portal hypertension in cystic fibrosis: A longitudinal prospective birth cohort study, Journal of Cystic Fibrosis, https://doi.org/10.1016/j.jcf.2019.09.016
ARTICLE IN PRESS
JID: JCF 4
[m5G;October 29, 2019;19:29]
C. Marco, F. Judith and W. Donna et al. / Journal of Cystic Fibrosis xxx (xxxx) xxx Table 4 Cox proprortional hazard hazard ratios for two or more elevated LFTs and development of portal hypertension, enzyme subgroup. ≥2 Elevated liver function tests
Unadjusted hazard ratios (95% CI)
Adjusted Hazard ratios (95% CI)∗
ALT Yes No p value
2.44 (1.33–4.47)1.00
2.69 (1.43–5.04)1.00
0.004
0.002
ALP Yes No p value
1.73 (0.95–3.17)1.00
1.86 (1.00–3.46)1.00
0.074
0.049
GGT Yes No p value
2.80 (1.19–6.86)1.00
5.71 (3.11–10.47)1.00
0.019
<0.001
∗
Stratified by clinic and adjusted for number of tests.
tions (ALT/ALP, ALT/GGT, GGT/ALP). One abnormal test only was not associated with PH for ALT (n = 87) (HR 1.55; 95% CI 0.79 to 3.03, p = 0.24) or ALP (n = 146) (HR 1.51; 95% CI: 0.83 to 2.74, p = 0.18) but was negatively associated with GGT (n = 304) (HR 0.37; 95% CI 0.20 to 0.67; p = 0.001). There was a higher risk of developing PH with combinations than with individual enzymes: ALT/ALP (n = 44) (HR 3.59; 95% CI, 1.76 to 7.31; p < 0.001); ALT/GGT (n = 39) (HR 5.56; 95% CI 2.82 to 10.98: p < 0.001) and ALP/GGT (n = 30) (HR 5.74 95% CI 2.78 to 11.86, p < 0.001) (see Table S4 for separate clinic results). Kaplan-Meier survival curves and the log rank test results for the 3 liver enzymes individually are depicted in Figure S1a – S1c and show the earlier onset of PH for those with 2 or more elevated enzymes on at least 2 occasions. Discussion This study was conducted on a large cohort of CF infants diagnosed by NBS and referred to one of the two participating CF centres which have attending gastroenterologists and provides accurate and reliable data on the occurrence and onset of PH as evident by the appearance of splenomegaly and confirmed by imaging with ultrasound or NMR. The study represents the first time a CF neonatal cohort has been followed to 18.5 years, assessing the occurrence and subsequent evolution of severe liver disease. The results highlight an overall prevalence of 8.8% in the combined clinic data (8.6% VCFC and 9.0% CHW) with incident rates of 2.81/10 0 0 patient years (3–7 years of age), 3.95/10 0 0 patient years (8–12 years of age) and 2.55/10 0 0 patient years (13–18 years of age) and a significantly higher (p < 0.001) mortality/transplant occurrence in those with severe liver disease with PH as compared with the non-PH subgroup. Multiple studies have assessed prevalence of cirrhosis with PH and a wide range has been found from 1% to 22% but these extreme values are likely underestimations or overestimations (respectively) and most of the larger studies have a prevalence in the 5–7% range. In this regard the most recent and one of the largest studies of CFPI patients is the French Gene Modifier Study (FGMS) [26] of severe liver disease (SLD) which reported a prevalence of 175/3328 (5.26%), 60% of that found in the current study. When comparing these two studies the French study used a less conservative definition of SLD that included a majority of patients (56%) without PH, and PH occurred in only 44% of the total 175 with SLD and therefore prevalence of PH was only 2.3% compared with 8.8% in the present study. The incidence rates would also be decreased accordingly. This marked difference could be explained by the fact that annual examination in France only started in 2001, 16 years after annual data collection commenced. The latter questions the
reliability of the data prior to 2001 and could contribute to the rising incidence in the 20–30 year old group after mandatory annual examination commenced, i.e. there was a delay in diagnosis until transition to adult centres or to the adult section of a CF centre. Obviously this would not account for the difference between FGMS and our study unless in the FGMS some patients developed severe lung disease and died during childhood, adolescence or early adulthood either prior to the development of PH or with unrecognized PH to account for the difference between their study and ours, but data on mortality and transplants were not provided to support or dispute this possibility. Combined mortality/transplant (MTx) data in the present study was over 4x higher (p < 0.001) in the PH versus the non-PH subgroup in accord with the match controlled Irish cohort study [27]. Of the 51 PH patients, those with MTx had a significantly lower age of onset of PH of 8 years compared to 12 years for those without MTx (p = 0.007) supporting the suggestion of other studies [2,11,13,15] that severe morbidity or mortality and aggressive liver disease occurs in patients with early onset PH i.e. it is a midchildhood phenomenon and de novo occurrence in adult patients is rare or non-existent [11,13,15,29]. The presence of PH is thus required to justify inclusion in the SLD category as PH is the marker of SLD used in several studies [2,3,13,27–29] including the International Gene Modifier study [2]. It is possible that PH and the requirement for liver transplant following liver failure may increase over 30 years of age. The above discussion has not accounted for the occurrence of Nodular Regenerative Hyperplasia (NRH) a recently described entity over the last 12 months [30]. They describe PH occurring in over 25 patients with CF and PH without liver cirrhosis on ex-plants but with PV occlusion of varying severity. We have not seen this entity on our ex-plant pathology but all CF centres should be aware of this phenomena to investigate its origin. The individual liver enzymes ALT and GGT and the combinations ALT/GGT, GGT/ALP and ALT/ALP if elevated on two or more occasions in early life were significantly predictive of the later development of PH. Whether the clinics were combined or analysed separately and regardless of the greater number of tests done in Verona, in general a similar pattern emerged in relation to abnormal LFTs and PH. As less than 30% of PH patients exhibited these patterns they were not useful for predicting PH in individual patients. However, the elevated enzyme patterns in earlier life were significantly associated with earlier onset of PH. These findings suggest that there are two emerging phenotypes in CFSLD: firstly a severe phenotype with early onset MBC with PH with complications including variceal bleeding and in some early liver failure necessitating liver transplantation, and secondly a milder phenotype with MBC and PH who have a more benign course with minimal
Please cite this article as: C. Marco, F. Judith and W. Donna et al., Occurrence, outcomes and predictors of portal hypertension in cystic fibrosis: A longitudinal prospective birth cohort study, Journal of Cystic Fibrosis, https://doi.org/10.1016/j.jcf.2019.09.016
JID: JCF
ARTICLE IN PRESS
[m5G;October 29, 2019;19:29]
C. Marco, F. Judith and W. Donna et al. / Journal of Cystic Fibrosis xxx (xxxx) xxx
LFT disturbance, less complications and minimal or no requirement for transplantation extending into adulthood until at least 30 years of age. Although this study was based on a total cohort collected in two clinics over 20 years, the small sample size in subgroups, such as gender and age group, has potentially resulted in a lack of statistical significance. Given median survival is now extending into the late 40’s a large multicentred international study needs to be performed to address the issues outlined above. Declaration of Competing Interest The authors declare that there are no conflicts of interest. Acknowledgements The authors would like to acknowledge Ms Cheryl Frazer and Mrs Margie Gruca for their tireless efforts in retrieving data, maintaining the database and formatting the manuscript for publication. Supplementary materials Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.jcf.2019.09.016. References [1] Sokol RJ, Durie PR. Recommendations for management of liver and biliary tract disease in cystic fibrosis. Cystic Fibrosis Foundation Hepatobiliary Disease Consensus Group. J Pediatr Gastroenterol Nutr 1999;28:S1–3. [2] Stonebraker JR, Ooi CY, Pace RG, Corvol H, Knowles MR, Durie PR, Ling SC. Features of severe liver disease with portal hypertension in patients with cystic fibrosis. Clin Gastroenterol Hepatol 2016;14:1207–15. [3] Flass T, Narkewicz MR. Cirrhosis and other liver disease in cystic fibrosis. J Cyst Fibrosis 2013;12:116–24. [4] Ling SC, Wilkinson JD, Hollman AS, McColl J, Evans TJ, Paton JY. The evolution of liver disease in cystic fibrosis. Arch Dis Child 1999;81:129–32. [5] Pandit C, Graham C, Selvadurai H, Gaskin K, Cooper P, Van Asperen P. Festival food coma in cystic fibrosis. Pediatric Pulmonol 2013;48:725–7. [6] Nightingale S, O’Loughlin EV, Dorney SF, et al. Isolated liver transplantation in children with cystic fibrosis – an Australian experience. Pediatr Transplant 2010;14:779–85. [7] Leung DH, Ye W, Molleston JP, et al. Baseline ultrasound and clinical correlates in children with cystic fibrosis. J Pediatr 2015;167:862–8. [8] Lenaerts C, Lapierre C, Patriquin H, et al. Surveillance for cystic fibrosis-associated hepatobiliary disease: early ultrasound changes and predisposing factors. J Pediatr 2003;143:343–50.
5
[9] Lamireau T, Monnereau S, Martin S, Marcotte J, Winnock M, Alvarez F. Epidemiology of liver disease in cystic fibrosis: a longitudinal study. J Hepatol 2004;41:920–5. [10] Fitzsimmons. Cystic fibrosis foundation patient registry. Elsevier; 1997. Annual data base report Bethesda, Maryland. [11] Scott-Jupp R, Lama M, Tanner MS. Prevalence of liver disease in cystic fibrosis. Arch Dis Child 1991;66:698–701. [12] Vawter GF, Shwachman H. Cystic fibrosis in adults, an autopsy study. Pathol Annual 1979;14:357–82. [13] Nash KL, Allison ME, McKeon D, et al. A single centre experience of liver disease in adults with cystic fibrosis 1995–2006. J Cyst Fibrosis 2008;7:252–7. [14] Feigelson J, Anagnostopoulos C, Poquet M, Pecau Y, Munck A, Navarro J. Liver cirrhosis in cystic fibrosis – therapeutic implications and long term follow up. Arch Dis Child 1993;68:653–7. [15] Colombo C, Battezzati PM, Crosignani A, Morabito A, Costantini D, Padoan R. Liver disease in cystic fibrosis: a prospective study on incidence, risk factors and outcome. Hepatology 2002;36:1374–82. [16] Corbett K, Kelleher S, Rowland M, et al. Cystic fibrosis-associated liver disease: a population-based study. J Pediatr 2004;145:327–32. [17] Mayer-Hamblett N, Kloster M, Ramsey BW, Narkewicz MR, Saiman L, Goss CH. Incidence and clinical significance of elevated liver function tests in cystic fibrosis clinical trials. Contemp Clin Trials 2013;34:232–8. [18] Wilschanski M, Rivlin J, Cohen S, et al. Clinical and genetic risk factors for cystic fibrosis-related liver disease. Pediatrics 1999;103:52–7. [19] Narkewicz MR. Markers of cystic fibrosis-associated liver disease. J Pediatr Gastroenterol Nutr 2001;32:421–2. [20] Cipolli M, Castellani C, Wilcken B, Massie J, McKay K, Gruca M, et al. Pancreatic phenotype in infants with cystic fibrosis identified by two neonatal mutational screening programs. Arch Dis Child 2007;92:842–6. [21] Waters DL, Dorney SFA, Gaskin KJ, Gruca MA, O’Halloran M, Wilcken B. Pancreatic function in infants identified as having cystic fibrosis in a neonatal screening program. New Engl J Med 1990;322:303–8. [22] Soldin SJ. Pediatric reference ranges. Soldin SJ, editor. second ed. Washington DC: AACC Press; 1997. [23] George J, Denney-Wilson E, Okely AD, Hardy LL, Aitken R. The population distributions, upper normal limits and correlations between liver tests among Australian adolescents. J Paediatr Child Health 2008;44:579–85. [24] Altman DG, Bland JM. Interaction revisited: the difference between two estimates. BMJ 2003;326(7382):219. [25] Norman GE, Streiner DL. Biostatistics: the bare essentials. St Louis: Mosby Year Book Inc.; 1994. p. 108–18. [26] Boelle P.-Y., Debray D., Guillot L., Clement A., Corvol H. Cystic fibrosis liver disease: outcomes and risk factors in a large cohort of French patients. doi:10. 1002/hep.30148. [27] Rowland M, Gallagher CG, O’Laoide R, et al. Outcome in cystic fibrosis liver disease. Am J Gastroenterol 2011;106:104–9. [28] Melzi ML, Kelly DA, Colombo C, Jara P, Manzanares J, Colledan M, et al. Liver transplant in cystic fibrosis: a poll among European centers. A study from the European Liver Transplant Registry. Transpl Int 2006;19:726–31. [29] Desmond CP, Wilson J, Bailey M, Clark D, Roberts SK. The benign course of liver disease in adults with cystic fibrosis and the effect of ursodeoxycholic acid. Liver Int 2007;27:1402–8. [30] Wu H, Vu M, Dhingra S, Ackah R, Goss JA, Rana A, Quintanilla N, Patel K, Leung DH. Obstructive portal venopathy without cirrhosis is prevalent in pediatric cystic fibrosis liver disease with portal hypertension. Clin Gastroenterol Hepatol 2019;17:2134–6.
Please cite this article as: C. Marco, F. Judith and W. Donna et al., Occurrence, outcomes and predictors of portal hypertension in cystic fibrosis: A longitudinal prospective birth cohort study, Journal of Cystic Fibrosis, https://doi.org/10.1016/j.jcf.2019.09.016