BACKGROUND AND OBJECTIVES

Cystic fibrosis (CF) screen–positive infants with an inconclusive diagnosis (CFSPID) are infants in whom sweat testing and genetic analysis does not resolve a CF diagnosis. Lack of knowledge about the health outcome of these children who require clinical follow-up challenges effective consultation. Early predictive biomarkers to delineate the CF risk would allow a more targeted approach to these children.

METHODS

Prospective, longitudinal, multicenter, Canada-wide cohort study of CF positive–screened newborns with 1 to 2 cystic fibrosis transmembrane conductance regulator gene variants, of which at least 1 is not known to be CF-causing and/or a sweat chloride between 30 and 59 mmol/L. These were monitored for conversion to a CF diagnosis, pulmonary, and nutritional outcomes.

RESULTS

The mean observation period was 7.7 (95% confidence interval 7.1 to 8.4) years. A CF diagnosis was established for 24 of the 115 children with CFSPID (21%) either because of reinterpretation of the cystic fibrosis transmembrane conductance regulator genotype or because of increase in sweat chloride concentration ≥60 mmol/L. An initial sweat chloride of ≥40 mmol/l predicted conversion to CF on the basis of sweat testing. The 91 remaining children with CFSPID were pancreatic sufficient and showed normal growth until school age. Pulmonary function as well as lung clearance index in a subgroup of children with CFSPID were similar to that of healthy controls.

CONCLUSIONS

Children with CFSPID have good nutritional and pulmonary outcomes at school age, but rates of reclassifying the diagnosis are high. The initial sweat chloride test can be used as a biomarker to predict the risk for CF in CFSPID.

What’s Known on This Subject:

Cystic fibrosis screen–positive infants with an inconclusive diagnosis are generally doing well in respect to their nutritional status and lung function until school age. Guidelines recommend continuous clinical monitoring until the age of 6 years.

What This Study Adds:

We compared outcome of cystic fibrosis screen–positive infants with an inconclusive diagnosis children to a parallel cohort of pancreatic sufficient and insufficient CF patients. We showed, for the first time, that the initial sweat test predicts conversion to CF during follow-up.

Newborn screening (NBS) for cystic fibrosis (CF) for early identification of children with a diagnosis of CF has been implemented in many countries around the world to mitigate the risk for malnutrition and early onset of lung disease.1  Despite development of CF NBS protocols that include 2 to 3 tiers of testing to balance sensitivity and specificity of the screening algorithm, high false positivity rates in screened newborns remain a burden to families receiving false-positive or inconclusive screening results as well as to the health care system.25 

Most of the CF NBS algorithms are based on initial testing for elevated immunoreactive trypsinogen (IRT), followed by screening for variants in the cystic fibrosis transmembrane conductance regulator (CFTR) gene by using the initial blood spot sample.1  A positive screen result triggers referral to a specialized CF center for sweat testing, which in most cases confirms or refutes a CF diagnosis.

Among the infants who screened positive for CF are those in whom either 2 CFTR gene variants are identified (of which at least 1 is not CF-causing, as per the CFTR2 database) and/or for whom consequent sweat testing led to a borderline sweat test result (sweat chloride concentration between 30 and 59 mmol/L). The term “cystic fibrosis screen–positive with inconclusive diagnosis” (CFSPID) or “CF-related metabolic syndrome” has been coined to direct special attention to this group of asymptomatic newborns, and clinical monitoring is recommended by international guidelines.69  The worry is that a diagnosis of CF cannot be absolutely dismissed because interpretation of the genetic variants of unknown or varying significance may change, sweat chloride concentration may change over time and may overlap with those measured in CF patients, and, finally, genetic and sweat test constellations resemble those observed in older patients with CFTR-related disorders, which is a clinical entity associated with CFTR dysfunction not fulfilling CF diagnostic criteria.7,1012 

In the first interim analysis of our longitudinal, prospective Canada-wide CF NBS study we have previously reported the health outcomes of children with CFSPID in early childhood at 2 to 3 years of age.13  In this study, we aim to evaluate the health outcomes of children with CFSPID at school age compared with CF patients identified in parallel by the Canadian CF NBS programs. Secondarily, we examined whether the sweat chloride, the sweat chloride pattern over time, or IRT can be used to predict risk for CF.

This is an ongoing longitudinal prospective Canada-wide CF NBS study which commenced in 2008 (CanCFSPID). The study was approved by research ethics boards (REBs) at the Hospital for Sick Children (1000012563) as the lead site and REBs at the University of British Columbia Children's and Women's Health Center, University of Calgary Conjoint Health, University of Alberta Health, University of Saskatchewan Biomedical, University of Manitoba Health, Western University, Children’s Hospital of Eastern Ontario, Queen's University Health Sciences & Affiliated Teaching Hospital, Izaak Walton Killam Hospital for Children, and Newfoundland and Labrador Health as collaborative sites. Informed consent was signed by the parent(s) or primary caregiver(s) of the participating children. The different provincial screening algorithms and total number of positive cases are summarized in Supplemental Table 4. Data for the CF NBS control group were obtained from the Canadian CF registry (REB#1000059950).

CFSPID

CFSPID was diagnosed on the basis of (1) a positive CF NBS result and (2) either a sweat chloride value <30 and 2 CFTR gene variants (mutations), at least 1 of which has unclear phenotypic consequences or an intermediate sweat chloride value (30–59 mmol/L) and <2 CF-causing gene variants (mutations).14 

CF Controls

Children with CF were diagnosed on the basis of (1) a positive CF NBS result and (2) the presence of 2 CF-causing CFTR gene variants and/or a sweat chloride concentration of ≥60 mmol/L. Children with CF were further divided into those who maintained pancreatic sufficiency and those who developed pancreatic insufficiency on the basis of fecal elastase measurements, with pancreatic sufficiency being determined with fecal elastase >200 microg/g stool,15  fecal fat measurements, and clinical assessments.

Diagnostic Tests

The CFTR genotype was obtained from NBS gene variant analysis and consequent CFTR gene exon sequencing at baseline. The CFTR2 database (status January 2021) was used to determine the pathogenicity of the CFTR gene variants.16  Sweat chloride was performed at baseline and then repeated every 6 months for the first 2 years and yearly thereafter.

To evaluate health outcome, anthropometric measures, such as weight, height, and BMI, were collected at baseline, then every 6 months until 2 years of age, and then yearly thereafter, and transformed into z scores for age by using the World Health Organization standard curves.17  All children at the age ≥5 years were asked to perform spirometry testing to measure the forced expiratory volume in 1 second (FEV1). FEV1 z scores were then calculated by using the previously published Macro, Version 1 (April 7, 2013).18  Multiple breath nitrogen washout measurements were performed at the lead site (Toronto) as previously described19  by using the Exhalyzer D (Eco Medics AG, Duernten, Switzerland). Lung clearance index (LCI) was calculated and reported if there were 2 technically acceptable trials. In addition, we performed oropharyngeal swabs in CFSPID and also induced or expectorated sputum and broncho-alveolar lavage for CF patients.

Descriptive summaries are provided as means and SD for normally distributed data and as median and interquartile range for nonparametric distributed data. Student t testing and an α of 0.05 was set as the significance level.

To address the primary objective of clinical outcome in CFSPID, weight and height z score trajectories as well as FEV1 z score trajectories were analyzed by using hierarchical linear regression modeling of the longitudinal change in anthropometric measures and pulmonary function over time with a random effect for age and subject. We plotted nonparametric trajectories using local regression smoothing and used piecewise regression to better capture the varying rates of change across age. Because we saw a peak in the curve fitting of the longitudinal growth data between the ages of 1 and 2 years, we used 1.5 years as the break point for the linear regression model.

To reduce the bias of differences in the frequency of sampling for microbiologic cultures, which we used as marker of lung health, we applied logistic regression to model the probability of a positive bacteriology result, with repeated tests across age, treating participants as a random effect. We capped the analysis of the clinical parameters at the age of 6 and 7 years to achieve consistent sample sizes across the ages and not overinterpret trajectories at the ages of 8 to 10 years, when numbers were lower.

To address the second objective of evaluating sweat chloride as potential predictors of the risk for CF, we modeled the time to first sweat ≥60 mmol/L in the subsample of participants whose first sweat chloride measurement was <60 mmol/L using a proportional hazards model. All patients were divided into those who met the criteria for a diagnosis of CF on the basis of CFTR genotype and those who did not, with the latter group further subdivided into those whose initial sweat was <40 mmol/L and those whose initial sweat was ≥40 mmol/L. This cutoff was identified as the value that maximized the χ2 statistic for the model. We then plotted the Kaplan-Meier estimates of the cumulative probability of a sweat chloride test ≥60 mmol/L across the 3 groups. We visually assessed the relationship between the IRT at birth and CF diagnosis on the basis of sweat ≥60 mmol/L using simple scatterplots of IRT versus age at the diagnostic sweat or the last available sweat.

Analysis was done by using SAS 9.4 (SAS Institute Inc, Cary, NC), and figures were done in R (version 4.0.2; R Foundation for Statistical Computing, Vienna, Austria).

Baseline characteristics as well as age of follow-up of the 3 diagnostic groups (CFSPID, pancreatic sufficient cystic fibrosis [CFPS], and pancreatic insufficient cystic fibrosis [CFPI]) are summarized in Table 1. A total of 115 children with CFSPID were included in the study (CFTR genotypes are provided in Supplemental Table 5), with a mean follow-up of 7.7 years at the time of this report. A CF diagnosis was established for 24 of the 115 children with CFSPID (21%) during clinical follow-up. For 12 of 24 children, the CF diagnosis was based on reinterpretation of their second CFTR gene variant as a CF-causing allele (L206W [6], R117C [3], D110H [1], F191V [1], and Q1476X [1]), and, for another 12 of 24 children with CFSPID, the CF diagnosis was based on an increase of the sweat chloride concentration to levels ≥60 mmol/L over time. A total of 91 CFSPID remained in this category. Of the 115 children with CFSPID, 27 were lost to follow-up, with data available until an average age of 3.1 years (range: 0.2–8.6 years). We compared the CFSPID cohort with CF patients identified by CF NBS during the same time period. These included 442 patients with CFPI and 104 patients with CFPS. The 24 CFSPID who converted to CF were added to the CFPS group for the clinical comparative analysis because we were interested in the outcome of “true” CFSPID. Serial fecal elastase measurements of children with CFSPID and CFPS as well as CFPS between the ages of 0 and 2.5 years confirmed pancreatic sufficiency. The initial sweat test and IRT were significantly different between the 3 diagnostic groups.

TABLE 1

Baseline Characteristics and Age to Follow-up of CFSPID, CFPS, and CFPI

CFSPID (n = 91)CFPS (n = 104)CFPI (n = 442)CFSPID and CFPS, PCFSPID and CFPI, P
Female, n (%) 51 (56) 43 (41) 236 (53) .040 .6 
Current age, mean (95% CI) 7.7 (7.1 to 8.4) 7.5 (6.8 to 8.1) 6.7 (6.4 to 7.0) .6 .007 
IRTa(95% CI) 88 (71 to  105)b 121 (101 to 140)c 188 (177 to 198)d .013 .0001 
Initial sweat chloride, mean (95% CI) 31.4 (27.6 to 35.2)b 64.4 (60.7 to 68.2)e 91.7 (89.9 to93.6)f <.0001 .0001 
CFSPID (n = 91)CFPS (n = 104)CFPI (n = 442)CFSPID and CFPS, PCFSPID and CFPI, P
Female, n (%) 51 (56) 43 (41) 236 (53) .040 .6 
Current age, mean (95% CI) 7.7 (7.1 to 8.4) 7.5 (6.8 to 8.1) 6.7 (6.4 to 7.0) .6 .007 
IRTa(95% CI) 88 (71 to  105)b 121 (101 to 140)c 188 (177 to 198)d .013 .0001 
Initial sweat chloride, mean (95% CI) 31.4 (27.6 to 35.2)b 64.4 (60.7 to 68.2)e 91.7 (89.9 to93.6)f <.0001 .0001 
a

IRT reflects the NBS-based measurements.

b

n = 85.

c

n = 63.

d

n = 230.

e

n = 87.

f

n = 340.

Anthropometric Measures

Serial fecal elastase measurements were available for 67 of 91 children with CFSPID, indicating pancreas sufficiency. Anthropometric measures at the age of 6 years revealed normal weight and height data for children with CFSPID (Table 2; Supplemental Table 4; Supplemental Fig 3). Similarly, children with CFPS showed normal weight and height at the age of 6 years. In contrast, the weight and height of children with CFPI at the age of 6 years was below healthy World Health Organization controls. Interestingly, whereas CFPI had significantly lower birth weight compared with CFPS and CFSPID, our data revealed that the deviation in weight gain between children with CF and children with CFSPID mainly occurred at the age of ≥18 months. After a period of catch-up weight and height in children with CFPI resulting in similar weight across the groups at 18 months of age, children with CFSPID continued to show normal weight gain, whereas weight gain in children with CFPI and CFPS deviated from the World Health Organization standard curve, with reduced weight gain at the ages of 18 months to 7 years.

TABLE 2

Weight and Height z Scores (95%CI) for CFSPID, CFPS, and Insufficient CF Patients

Slope to Ages 0–18 Mo (95% CI)PEstimate at Age 18 Mo (95% CI)PSlope to Ages ≥18 Mo (95% CI)PEstimates at Age 6 y
Weight        
 CFSPID, n = 91 0.64 (0.42 to 0.86) <.0001 0.48 (0.25 to 0.71) — −0.02 (−0.08 to 0.04) .4 0.40 (0.16 to 0.64) 
 CFPS, n = 104 0.45 (0.25 to 0.65) <.0001 0.39 (0.18 to 0.60) — −0.08 (−0.13 to −0.02) .004 0.05 (−0.16 to 0.25) 
 CFPI, n = 405 0.86 (0.77 to 0.95) <.0001 0.25 (0.14 to 0.35) — −0.10 (−0.13 to −0.08) <.0001 −0.22 (−0.33 to −0.11) 
 Difference F = 7.51 .0006 — — F = 3.27 .038 — 
  CFSPID and CFPS 0.19 (−0.10 to 0.49) .2 0.09 (−0.22 to 0.40) .6 0.06 (−0.02 to 0.14) .2 — 
  CFSPID and CFPI −0.22 (−0.46 to 0.02) .071 0.23 (−0.02 to 0.49) .067 0.09 (0.02 to 0.15) .012 — 
  CFPS and CFPI −0.41 (−0.63 to −0.19) .0002 0.15 (−0.09 to 0.38) .2 0.03 (−0.03 to 0.09) .4 — 
Height        
 CFSPID, n = 91 0.06 (−0.18 to 0.30) .6 −0.03 (−0.25 to 0.19) — — — 0.01 (−0.21 to 0.24) 
 CFPS, n = 103 −0.08 (−0.30 to 0.13) .4 −0.12 (−0.32 to 0.07) — 0.01 (−0.01 to 0.03)1  .2 −0.12 (−0.32 to 0.08) 
 CFPI, n = 405 0.26 (0.16 to 0.36) <.0001 −0.46 (−0.56 to −0.37) — — — −0.28 (−0.39 to −0.18) 
 Difference F = 4.62 .010 — — n.s. n.s. — 
  CFSPID and CFPS 0.14 (−0.18 to 0.47) .4 0.13 (−0.16 to 0.42) .4 — — — 
  CFSPID and CFPI −0.20 (−0.46 to 0.06) .1 0.30 (0.06 to 0.54) .014 — — — 
  CFPS and CFPI −0.35 (−0.59 to −0.11) .005 0.17 (−0.05 to 0.39) .1 — — — 
Slope to Ages 0–18 Mo (95% CI)PEstimate at Age 18 Mo (95% CI)PSlope to Ages ≥18 Mo (95% CI)PEstimates at Age 6 y
Weight        
 CFSPID, n = 91 0.64 (0.42 to 0.86) <.0001 0.48 (0.25 to 0.71) — −0.02 (−0.08 to 0.04) .4 0.40 (0.16 to 0.64) 
 CFPS, n = 104 0.45 (0.25 to 0.65) <.0001 0.39 (0.18 to 0.60) — −0.08 (−0.13 to −0.02) .004 0.05 (−0.16 to 0.25) 
 CFPI, n = 405 0.86 (0.77 to 0.95) <.0001 0.25 (0.14 to 0.35) — −0.10 (−0.13 to −0.08) <.0001 −0.22 (−0.33 to −0.11) 
 Difference F = 7.51 .0006 — — F = 3.27 .038 — 
  CFSPID and CFPS 0.19 (−0.10 to 0.49) .2 0.09 (−0.22 to 0.40) .6 0.06 (−0.02 to 0.14) .2 — 
  CFSPID and CFPI −0.22 (−0.46 to 0.02) .071 0.23 (−0.02 to 0.49) .067 0.09 (0.02 to 0.15) .012 — 
  CFPS and CFPI −0.41 (−0.63 to −0.19) .0002 0.15 (−0.09 to 0.38) .2 0.03 (−0.03 to 0.09) .4 — 
Height        
 CFSPID, n = 91 0.06 (−0.18 to 0.30) .6 −0.03 (−0.25 to 0.19) — — — 0.01 (−0.21 to 0.24) 
 CFPS, n = 103 −0.08 (−0.30 to 0.13) .4 −0.12 (−0.32 to 0.07) — 0.01 (−0.01 to 0.03)1  .2 −0.12 (−0.32 to 0.08) 
 CFPI, n = 405 0.26 (0.16 to 0.36) <.0001 −0.46 (−0.56 to −0.37) — — — −0.28 (−0.39 to −0.18) 
 Difference F = 4.62 .010 — — n.s. n.s. — 
  CFSPID and CFPS 0.14 (−0.18 to 0.47) .4 0.13 (−0.16 to 0.42) .4 — — — 
  CFSPID and CFPI −0.20 (−0.46 to 0.06) .1 0.30 (0.06 to 0.54) .014 — — — 
  CFPS and CFPI −0.35 (−0.59 to −0.11) .005 0.17 (−0.05 to 0.39) .1 — — — 

Anthropometric measures are provided as World Health Organization z score as well as age-dependent changes of World Health Organization z scores divided by development before and after 18 mo of age by using hierarchical piecewise linear regression modeling of the longitudinal data, with a breakpoint at 18 mo.1  There was no significant difference in the slope for height z scores across the 3 groups after the age of 1.5 years; the estimate of the slope is pooled across the 3 groups. Wt and height z scores (95%CI) predicted at the age of 6 y are derived from the piecewise regression models. n.s., not significant; —, not applicable.

Lung Disease

During the time of follow-up, 3 children with CFSPID were treated for bronchitis, and 8 were treated for pneumonia with antibiotic therapy. Children with CFSPID (n = 38), for whom reliable lung function tests were available, showed normal FEV1, compared with that of healthy control norms, with an average FEV1 z score of 0.18 (95% confidence interval [CI] −0.21 to 0.56), which was not significantly different from the FEV1 measured in CFPS (n = 44; P = .7) and CFPI patients (n = 169, P = .1; Supplemental Fig 4). Because FEV1 is not sufficiently sensitive for detecting early lung disease, we also measured LCI in 17 children with CFSPID at one of the centers. The average LCI in CFSPID was 6.92 (95% CI 6.62 to 7.22), at a mean age of 7.99 (95% CI 6.88 to 9.1) years, which was not significantly different from that of healthy controls (7.1 [95% CI 6.92 to 7.28]; P = .86) but lower, compared with the average LCI in CF patients (9.13 [95% CI 8.64 to 9.61]; P < .0001; Fig 1). We also evaluated for presence of bacteria in the respiratory tract, taking into account the difference in the sampling rate between children with CFSPID (yearly) and CF patients (quarterly). We identified Methicillin-resistant Staphylococcus aureus (MRSA), Pseudomonas aeruginosa (PA), and Stenotrophomonas maltophilia (Steno) in children with CFSPID, with a frequency of 0.5%, 2.5%, and 2.0% between the ages of 0 and 7 years (Table 3). CFPS and CFPI patients had higher odds for a positive S aureus culture result, compared with that of children with CFSPID, but children with CFSPID had the same odds for a positive result for MRSA and Steno, compared with children with CF. CFPI revealed a trend toward a higher odds for a positive culture result with PA, compared with that of children with CFSPID.

FIGURE 1

Cross-sectional LCI2.5 results, comparing CFSPID, healthy controls, and CF patients. The graph reveals cross sectional LCI2.5 data from individual subjects, comparing most recent available LCI2.5 measurements in CFSPID children to a group of healthy controls and CF patients. The dotted line indicates upper limits of normal at 7.9.34  The average LCI2.5 in CFSPID (n = 17) was 6.92 (95% CI 6.62 to 7.22), 7.10 (95% CI 6.92 to 7.28) in 52 HC19  and 9.13 (95% CI 8.64 to 9.61) in 51 CF patients. Analysis of variance and Tukey’s multiple comparison test revealed statistically significant differences between CFSPID and CF (P < .0001) and no significant difference between CFSPID and HC (P = .86). The mean age of the groups were as follows CFSPID (7.99 [95% CI 6.88 to 9.1] years); HC (4.96 [95% CI 4.7 to 5.23] years); and CF (5.27 [95% CI 4.96 to 5.57] years).

FIGURE 1

Cross-sectional LCI2.5 results, comparing CFSPID, healthy controls, and CF patients. The graph reveals cross sectional LCI2.5 data from individual subjects, comparing most recent available LCI2.5 measurements in CFSPID children to a group of healthy controls and CF patients. The dotted line indicates upper limits of normal at 7.9.34  The average LCI2.5 in CFSPID (n = 17) was 6.92 (95% CI 6.62 to 7.22), 7.10 (95% CI 6.92 to 7.28) in 52 HC19  and 9.13 (95% CI 8.64 to 9.61) in 51 CF patients. Analysis of variance and Tukey’s multiple comparison test revealed statistically significant differences between CFSPID and CF (P < .0001) and no significant difference between CFSPID and HC (P = .86). The mean age of the groups were as follows CFSPID (7.99 [95% CI 6.88 to 9.1] years); HC (4.96 [95% CI 4.7 to 5.23] years); and CF (5.27 [95% CI 4.96 to 5.57] years).

Close modal
TABLE 3

Microbiology Results

MRSAStaph1 PaSteno
Probability of a positive sample result, % (95% CI)     
 CFSPID, n = 82 0.5 (0.1 to 3.4) 2.7 (2.2 to 3.l) 2.5 (1.1 to 5.2) 2.0 (1.0 to 4.3) 
 CFPS, n = 26 1.3 (0.3 to 5.9) 2.9 (2.2 to 3.8) 2.9 (1.2 to 6.9) 0.8 (0.2 to 3.4) 
 CFPI, n = 74 2.2 (1.3 to 3.6) 3.9 (3.5 to 4.33) 5.1 (3.9 to 6.7) 1.0 (0.6 to 1.7) 
CFPI versus CFSPID, odds ratio (95% CI); P 4.09 (.59 to 28.49); .2 1.69 (1.19 to 2.41); .004 2.14 (.93 to 4.94); .07 .47 (.18 to 1.21); .1 
CFPI versus CFPS, odds ratio (95% CI); P 1.76 (.33 to 9.50); .5 1.54 (1.00 to 2.35); .049 1.80 (.70 to 4.65); .2 1.28 (.25 to 6.50); .8 
CFPS versus CFSPID, odds ratio (95% CI); P 2.32 (.20 to 27.29); .5 1.10 (.66 to 1.83); .7 1.19 (.36 to 3.95); .8 .37 (.07 to 2.03); .2 
MRSAStaph1 PaSteno
Probability of a positive sample result, % (95% CI)     
 CFSPID, n = 82 0.5 (0.1 to 3.4) 2.7 (2.2 to 3.l) 2.5 (1.1 to 5.2) 2.0 (1.0 to 4.3) 
 CFPS, n = 26 1.3 (0.3 to 5.9) 2.9 (2.2 to 3.8) 2.9 (1.2 to 6.9) 0.8 (0.2 to 3.4) 
 CFPI, n = 74 2.2 (1.3 to 3.6) 3.9 (3.5 to 4.33) 5.1 (3.9 to 6.7) 1.0 (0.6 to 1.7) 
CFPI versus CFSPID, odds ratio (95% CI); P 4.09 (.59 to 28.49); .2 1.69 (1.19 to 2.41); .004 2.14 (.93 to 4.94); .07 .47 (.18 to 1.21); .1 
CFPI versus CFPS, odds ratio (95% CI); P 1.76 (.33 to 9.50); .5 1.54 (1.00 to 2.35); .049 1.80 (.70 to 4.65); .2 1.28 (.25 to 6.50); .8 
CFPS versus CFSPID, odds ratio (95% CI); P 2.32 (.20 to 27.29); .5 1.10 (.66 to 1.83); .7 1.19 (.36 to 3.95); .8 .37 (.07 to 2.03); .2 

Microbiology results are presented as the probability for a positive result accounting for all performed samples per age and per diagnostic group. This was done to account for the different sample rates between CFSPID and CF patients.1  The probability of a positive sample result increased with age (odds ratio for 1 year of age: 1.18; 95% CI 1.13 to 1.24; P < .0001). The probability of a positive sample result is shown at 3.5 y of age, the midpoint of the age range. However, the odds ratios do not vary with age.

A total of 12 of 103 (12%) children with CFSPID met CF diagnostic criteria during the follow-up period, on the basis of an increase of sweat chloride concentrations to values ≥60 mmol/l. We, therefore, asked whether early biomarkers, such as the first sweat chloride test, can predict a conversion to a CF diagnosis. Our analysis revealed that children with CFSPID with an initial sweat test of ≥40 mmol/L had a 10 times higher hazard of having a CF-converting sweat test of ≥60 mmol/l later in life, compared with those with an initial sweat of <40 mmol/L (hazard ratio: 12.1 [95% CI 2.6 to 55.6]; P = .001; Fig 2A). Of the 10 children with CFSPID with a sweat ≥40 mmol/L, 9 CFSPID CF (sweat) converted to a CF diagnosis between the ages of 2.8 and 4.4 years.

FIGURE 2

Early predictive biomarkers for CFSPID-CF conversion. A, Initial sweat test. This graph reveals the Cox proportional hazard model for an individual subject converting to a CF diagnosis on the basis of sweat ≥60 mmol/l. Lines represent cumulative probabilities, and shaded areas represent the 95% CI. The red line presents patients who are diagnosed as CF on the basis of a CF-causing CFTR genotype (CF by genotype) but for whom the initial sweat was <60 mmol/L. Their cumulative probability of a sweat ≥60 mmol/l was 0.77 (95% CI 0.57 to 0.92). The yellow and green line represents children with CFSPID at birth who do not have a CF-causing CFTR genotype (CFSPID at birth). Those with an initial sweat ≥40 mmol/l (n = 30; yellow line) have a cumulative probability for a sweat ≥60 mmol/l and thus converting to a CF diagnosis of 0.46 (95% CI 0.27 to 0.70), whereas those with an initial sweat of <40 mmol/l have a cumulative probability of 0.01 (95% CI 0.00 to 0.10). The hazard ratio are as follows: CF versus CFSPID ≥40 mmol/l (4.9 [95% CI 2.2 to 11.2]; P = .0001); CFSPID ≥40 mmol/L versus CFSPID <40 mmol/l (12.1 [95% CI 2.6 to 55.6]; P = .001). B and C, These graphs reveal all children with CFSPID and CF patients with first sweat <60 mmol and their IRT at birth. Solid circles represent the age at which the first sweat ≥60 mmol/l occurred. Open circles represent the age of the last sweat plus 1 year (assumes that the value of the last sweat is representative of their sweat level for 1 year). B, IRT: CSFPID at birth. Reveals CFSPID patients who did not have CF-causing CFTR genotypes. C, IRT: CF by genotype. Reveals those patients who meet diagnostic criteria for CF on the basis of a CFTR genotype but for whom the initial sweat was <60 mmol/l. We did not identify any IRT cutoff level to predict a later CF diagnosis in children with CFSPID on the basis of sweat ≥60 mmol/l. However, in patients with a CF diagnostic CFTR genotype but initial sweat <60 mmol/L, we observed a trend toward a higher risk of a sweat conversion to ≥60 mmol/l with higher NBS IRT values (P = .08) with those above an IRT level of 130 ng/mL having a 4.56 (95% CI 1.13 to 18.4; P = .033) greater hazard of having a sweat ≥60 mmol/l.

FIGURE 2

Early predictive biomarkers for CFSPID-CF conversion. A, Initial sweat test. This graph reveals the Cox proportional hazard model for an individual subject converting to a CF diagnosis on the basis of sweat ≥60 mmol/l. Lines represent cumulative probabilities, and shaded areas represent the 95% CI. The red line presents patients who are diagnosed as CF on the basis of a CF-causing CFTR genotype (CF by genotype) but for whom the initial sweat was <60 mmol/L. Their cumulative probability of a sweat ≥60 mmol/l was 0.77 (95% CI 0.57 to 0.92). The yellow and green line represents children with CFSPID at birth who do not have a CF-causing CFTR genotype (CFSPID at birth). Those with an initial sweat ≥40 mmol/l (n = 30; yellow line) have a cumulative probability for a sweat ≥60 mmol/l and thus converting to a CF diagnosis of 0.46 (95% CI 0.27 to 0.70), whereas those with an initial sweat of <40 mmol/l have a cumulative probability of 0.01 (95% CI 0.00 to 0.10). The hazard ratio are as follows: CF versus CFSPID ≥40 mmol/l (4.9 [95% CI 2.2 to 11.2]; P = .0001); CFSPID ≥40 mmol/L versus CFSPID <40 mmol/l (12.1 [95% CI 2.6 to 55.6]; P = .001). B and C, These graphs reveal all children with CFSPID and CF patients with first sweat <60 mmol and their IRT at birth. Solid circles represent the age at which the first sweat ≥60 mmol/l occurred. Open circles represent the age of the last sweat plus 1 year (assumes that the value of the last sweat is representative of their sweat level for 1 year). B, IRT: CSFPID at birth. Reveals CFSPID patients who did not have CF-causing CFTR genotypes. C, IRT: CF by genotype. Reveals those patients who meet diagnostic criteria for CF on the basis of a CFTR genotype but for whom the initial sweat was <60 mmol/l. We did not identify any IRT cutoff level to predict a later CF diagnosis in children with CFSPID on the basis of sweat ≥60 mmol/l. However, in patients with a CF diagnostic CFTR genotype but initial sweat <60 mmol/L, we observed a trend toward a higher risk of a sweat conversion to ≥60 mmol/l with higher NBS IRT values (P = .08) with those above an IRT level of 130 ng/mL having a 4.56 (95% CI 1.13 to 18.4; P = .033) greater hazard of having a sweat ≥60 mmol/l.

Close modal

We did not identify any IRT cutoff level to predict a later CF diagnosis in CFSPID patients on the basis of sweat ≥60 mmol/l (Fig 2B). However, in patients with a CF diagnostic CFTR genotype but initial sweat <60 mmol/L, we observed a trend toward a higher risk of a sweat conversion to ≥60 mmol/l with higher NBS IRT values (P = .08), with those above an IRT level of 130 ng/mL having a 4.56 (95% CI 1.13 to 18.4; P = .033) greater hazard of having a sweat ≥60 mmol/L (Fig 2C), as determined in a post hoc analysis.

Our prospective longitudinal study revealed that overall, the nutritional and pulmonary status of children with CFSPID in their school-aged years compares with healthy control children. However, 21% of the children with CFSPID met CF diagnostic criteria during the course of the follow-up. For those 11% who transitioned to CF on the basis of an increase in sweat chloride ≥60 mmol/l, the risk can be predicted by an initial sweat chloride test ≥40 mmol/L. For those who transition to CF on the basis of reinterpretation of CFTR gene variants, the risk of converting to a CF diagnostic sweat test might be predicted by high initial IRT.

A sweat chloride of ≥40 mmol/L on the initial sweat test after birth identifies children with CFSPID children as at risk for a future CF diagnosis. Although this observation needs to be validated in an independent cohort before it can be clinically implemented,20,21  our study is nevertheless the first to reveal that the initial sweat chloride test can be used as a predictive tool in the care of children with CFSPID. It is interesting to note that also children beyond the age of 2 years showed an increase in sweat chloride to CF diagnostic ranges, which may be due to ongoing maturation of the sweat gland, changes in the innervation or hormonal levels, or other yet to be identified factors. Because we propose that CFTR-related disease may not develop until adolescence and young adulthood, it is nevertheless too early to comment on the predictability of the initial sweat test for CFTR-related disease. The conversion rate of 21% is the same as reported previously from this then younger and also smaller cohort.13  It included those for whom a CF diagnosis was established on reinterpretation of the CFTR gene variants. Munck et al22  reported a higher CF conversion rate in their CFSPID cohort, mainly due to a larger portion of children with CFSPID changing to CF because of reinterpretation of CFTR genetic variants in their earlier recruited cohort. For these children, the initial IRT appears to be a good predictor for the confirmation of such a CF diagnosis because high IRT was associated with sweat conversion to ≥60 mmol/L. Although this finding needs to be confirmed in another cohort, this knowledge may be helpful in future clinical scenarios in which the clinical or functional consequence of a particular CFTR gene variant is not yet known. IRT as a predictive marker for a CF diagnosis in children with CFSPID was suggested by our group earlier.23 

Because all CFSPID were pancreatic sufficient, clinical parameters were not only compared with CFPI patients but, specifically, with CFPS patients who generally experience a milder CF disease course. This is the first study paying attention to this subclassification of CF NBS patients, which allows a more accurate risk assessment of the children with CFSPID. Comparing clinical parameters of CFSPID, CFPS, and CFPI patients next to each other displays the clinical spectrum of CF disease associated with different levels of CFTR function. This is not only helpful for diagnostic purposes but may also provide useful information to monitor CFTR modulator drug efficacy in young children with CF in the future. It also illustrates the similarity of children with CFSPID, compared with those classified as CFPS, challenging early dismissal from clinical monitoring because of a lack of symptoms in the young age group.

Children with CFSPID showed normal weight gain and growth until school age, which has been shown in other studies as well22,24  suggesting that their mild CFTR dysfunction had no significant effect on nutrient absorption as well as on energy metabolism. The observed increase in weight gain over the first 18 months in our study cohort compared with that of the World Health Organization cohort is consistent with a recently reported growth deviation of Canadian infants and toddlers from the World Health Organization percentiles, possibly due to differences in the underlying reference population.25  However, despite this slight deviation, our anthropometric measurements performed in children with CF are in agreement with those recently reported in the Baby Observational and Nutritional Study,26  allowing generalization of our observation. Compared with the US CF NBS cohort, we equally observed low birth weight in the Canadian infants with CFPI, with consequent weight gain and increase in height to the age of 18 months. Interestingly, at 18 months of age this positive nutritional trend in CF changed to a decline in weight gain until school age. Thus, although CF NBS prevents early malnutrition in CF by mitigating effects of maldigestion,27  it appears to not sufficiently protect from weight loss beyond early childhood.

Children with CFSPID had normal lung function at school age, demonstrated by normal FEV1 and LCI, suggesting an absence of lung disease. This corresponds to reports from other CFSPID cohorts.22,28  In comparison, more than one-half of the school-aged children with CF showed increased LCI results reflecting underlying CF lung disease. We were nevertheless surprised by the significant number of children with CFSPID in whom we found CF-related airway pathogens. Because only a few studies evaluated airway colonization of healthy control children, it is difficult to understand the clinical relevance of this finding.2932  In other cohort studies, reseachers also reported PA colonization in their children with CFSPID, ranging from 15% to 30%.11,22,24  Some children with CFSPID in France were treated with azithromycin and Dornase α,22  and Canadian children with CFSPID positive for PA underwent eradication therapy with inhaled tobramycin. However, at this point, it is unknown whether children with CFSPID with positive microbiologic culture results are at an increased risk for developing lung disease later on and whether they too should be subjected to preventive therapies.

Although this prospective longitudinal study benefited from a unique population-based and prospective approach, we acknowledge the absence of a parallel healthy or heterozygote control group (eg, those CF positive–screened newborns discharged as healthy carriers of the disease). This limited our ability to interpret the clinical significance of some findings in our CFSPID cohort. Another limitation is the use of 2 different biochemical assays to measure IRT in Canada’s West and Eastern provinces, which challenges comparison of absolute IRT levels as outcome markers. We acknowledge that several of our children with CFSPID carried one R117H 7T allele, which, in other jurisdictions, would not lead to a positive CF NBS.33 

In summary, we showed that school-aged children with CFSPID had good nutritional and pulmonary outcomes. However, 21% of the children with CFSPID are at risk for a later CF diagnosis, which can be predicted by the initial sweat test for those who convert to CF on the basis of sweat chloride. Future follow-up studies need to reveal whether an initial sweat chloride of ≥40 mmol/L not only predicts CFSPID at risk for CF but also those at risk for CFTR-related disease, consequently allowing stratification of children with CFSPID children into those requiring clinical follow-up and those who can be discharged. Ideally these children can be included in CF patient registries to facilitate collection of outcome data.

The CanCFSPID Study Group participants were as follows: Louise Taylor, Karen Tam, Abbie Watts-Dickens, Julie Avolio, Anne Smith, Jane Corbeil, Janine Forbes, Jane Mayo, Vanessa McMahon, Amy Schellenberg, Paula Barrett, Margaret Surette, Adriana Breen, Monica Dawe, Lindsay Clark, Shannon Gregory, and Angela MacDonald.

We thank Pranesh Chakraborty and Karen Tam from the Ontario CF NBS program for reviewing the article. We also thank all the participating children and families.

Dr Gonska conceptualized and designed the study, conducted the initial analysis, drafted the initial manuscript, and reviewed and revised the manuscript; Dr Ratjen has participated in the concept and design of the study, participated in the analysis and interpretation, drafted the initial manuscript, and reviewed and revised the manuscript; Drs Chilvers, Bjornson, Brusky, Kherani, Kosteniuk, Price, Morgan, Mateos-Corral, Hughes, Smith, Iqbal, Reismann, van Wylick, Derynck, and Solomon have participated in the concept and design of the study, coordinated and supervised data collection, participated in the analysis and interpretation, and reviewed and revised the manuscript; Dr Dupuis has participated in the concept and design of this study, conducted the initial analysis, drafted the initial manuscript, and reviewed and revised the manuscript; Ms Keenan has participated in the concept and design of the study, participated in the analysis and interpretation, designed the data collection instruments, contributed to the initial analysis, and reviewed and revised the manuscript; Mr Au has participated in the concept and design of the study, has participated in the analysis and interpretation, contributed to the initial analysis, and reviewed and revised the manuscript; Ms Burgess, Fairservice, Jober, Ittermann, Donnelly, Arpin, Hammel, and Henderson have participated in the concept and design of the study, participated in the analysis and interpretation, collected data, and reviewed and revised the manuscript; and all authors approved the final manuscript as submitted.

FUNDING: No external support.

COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2021-052822.

CF

cystic fibrosis

CFPI

pancreatic insufficient cystic fibrosis

CFSPID

cystic fibrosis screen–positive infants with an inconclusive diagnosis

CFPS

pancreatic sufficient cystic fibrosis

CI

confidence interval

FEV1

forced expiratory volume in 1 second

IRT

immunoreactive trypsinogen

LCI

lung clearance index

MRSA

methicillin-resistant Staphylococcus aureus

NBS

newborn screening

PA

Pseudomonas aeruginosa

REB

research ethics board

Steno

Stenotrophomonas maltophilia

1
Gonska
T
,
Ratjen
F
.
Newborn screening for cystic fibrosis
.
Expert Rev Respir Med
.
2015
;
9
(
5
):
619
631
2
Hayeems
RZ
,
Miller
FA
,
Barg
CJ
, et al
.
Psychosocial response to uncertain newborn screening results for cystic fibrosis
.
J Pediatr
.
2017
;
184
:
165
171.e1
3
Hayeems
RZ
,
Miller
FA
,
Vermeulen
M
, et al
.
False-positive newborn screening for cystic fibrosis and health care use
.
Pediatrics
.
2017
;
140
(
5
):
e20170604
4
Perobelli
S
,
Zanolla
L
,
Tamanini
A
,
Rizzotti
P
,
Maurice Assael
B
,
Castellani
C
.
Inconclusive cystic fibrosis neonatal screening results: long-term psychosocial effects on parents
.
Acta Paediatr
.
2009
;
98
(
12
):
1927
1934
5
Johnson
F
,
Southern
KW
,
Ulph
F
.
Psychological impact on parents of an inconclusive diagnosis following newborn bloodspot screening for cystic fibrosis: a qualitative study
.
Int J Neonatal Screen
.
2019
;
5
(
2
):
23
6
Munck
A
,
Mayell
SJ
,
Winters
V
, et al;
ECFS Neonatal Screening Working Group
.
Cystic fibrosis screen positive, inconclusive diagnosis (CFSPID): a new designation and management recommendations for infants with an inconclusive diagnosis following newborn screening
.
J Cyst Fibros
.
2015
;
14
(
6
):
706
713
7
Barben
J
,
Castellani
C
,
Munck
A
, et al;
European CF Society Neonatal Screening Working Group (ECFS NSWG)
.
Updated guidance on the management of children with cystic fibrosis transmembrane conductance regulator-related metabolic syndrome/cystic fibrosis screen positive, inconclusive diagnosis (CRMS/CFSPID)
.
J Cyst Fibros
.
2021
;
20
(
5
):
810
819
8
Borowitz
D
,
Parad
RB
,
Sharp
JK
, et al;
Cystic Fibrosis Foundation
.
Cystic Fibrosis Foundation practice guidelines for the management of infants with cystic fibrosis transmembrane conductance regulator-related metabolic syndrome during the first two years of life and beyond
.
J Pediatr
.
2009
;
155
(
6 Suppl
):
S106
S116
9
Ren
CL
,
Borowitz
DS
,
Gonska
T
, et al
.
Cystic fibrosis transmembrane conductance regulator-related metabolic syndrome and cystic fibrosis screen positive, inconclusive diagnosis
.
J Pediatr
2017
;
181S
:
S45
S51.e1
10
Keenan
K
,
Dupuis
A
,
Griffin
K
,
Castellani
C
,
Tullis
E
,
Gonska
T
.
Phenotypic spectrum of patients with cystic fibrosis and cystic fibrosis-related disease carrying p.Arg117His
.
J Cyst Fibros
.
2019
;
18
(
2
):
265
270
11
Castaldo
A
,
Cimbalo
C
,
Castaldo
RJ
, et al
.
Cystic fibrosis-screening positive inconclusive diagnosis: newborn screening and long-term follow-up permits to early identify patients with CFTR-related disorders
.
Diagnostics (Basel)
.
2020
;
10
(
8
):
570
12
Bombieri
C
,
Claustres
M
,
De Boeck
K
, et al
.
Recommendations for the classification of diseases as CFTR-related disorders
.
J Cyst Fibros
.
2011
;
10
(
suppl 2
):
S86
S102
13
Ooi
CY
,
Castellani
C
,
Keenan
K
, et al
.
Inconclusive diagnosis of cystic fibrosis after newborn screening
.
Pediatrics
.
2015
;
135
(
6
):
e1377
e1385
14
Southern
KW
,
Barben
J
,
Gartner
S
, et al
.
Inconclusive diagnosis after a positive newborn bloodspot screening result for cystic fibrosis; clarification of the harmonised international definition
.
J Cyst Fibros
.
2019
;
18
(
6
):
778
780
15
Beharry
S
,
Ellis
L
,
Corey
M
,
Marcon
M
,
Durie
P
.
How useful is fecal pancreatic elastase 1 as a marker of exocrine pancreatic disease?
J Pediatr
.
2002
;
141
(
1
):
84
90
16
McCague
AF
,
Raraigh
KS
,
Pellicore
MJ
, et al
.
Correlating cystic fibrosis transmembrane conductance regulator function with clinical features to inform precision treatment of cystic fibrosis
.
Am J Respir Crit Care Med
.
2019
;
199
(
9
):
1116
1126
17
WHO Multicentre Growth Reference Study Group
.
WHO Child Growth Standards based on length/height, weight and age
.
Acta Paediatr Suppl
.
2006
;
450
:
76
85
18
Quanjer
PH
,
Stanojevic
S
,
Cole
TJ
, et al;
ERS Global Lung Function Initiative
.
Multi-ethnic reference values for spirometry for the 3-95-yr age range: the global lung function 2012 equations
.
Eur Respir J
.
2012
;
40
(
6
):
1324
1343
19
Stanojevic
S
,
Davis
SD
,
Retsch-Bogart
G
, et al
.
Progression of lung disease in preschool patients with cystic fibrosis
.
Am J Respir Crit Care Med
.
2017
;
195
(
9
):
1216
1225
20
Clinical prediction rules
. In:
Guyatt
GRD
,
Meade
MO
,
Cook
DJ
, eds.
Users’ Guides to the Medical Literature
.
New York, NY
:
McGraw-Hill
;
2014
21
Alba
AC
,
Agoritsas
T
,
Walsh
M
, et al
.
Discrimination and calibration of clinical prediction models: users’ guides to the medical literature
.
JAMA
.
2017
;
318
(
14
):
1377
1384
22
Munck
A
,
Bourmaud
A
,
Bellon
G
,
Picq
P
,
Farrell
PM
;
DPAM Study Group
.
Phenotype of children with inconclusive cystic fibrosis diagnosis after newborn screening
.
Pediatr Pulmonol
.
2020
;
55
(
4
):
918
928
23
Ooi
CY
,
Sutherland
R
,
Castellani
C
, et al
.
Immunoreactive trypsinogen levels in newborn screened infants with an inconclusive diagnosis of cystic fibrosis
.
BMC Pediatr
.
2019
;
19
(
1
):
369
24
Salinas
DB
,
Sosnay
PR
,
Azen
C
, et al
.
Benign outcome among positive cystic fibrosis newborn screen children with non-CF-causing variants
.
J Cyst Fibros
.
2015
;
14
(
6
):
714
719
25
Park
AL
,
Tu
K
,
Ray
JG
;
Canadian Curves Consortium
.
Differences in growth of Canadian children compared to the WHO 2006 Child Growth Standards
.
Paediatr Perinat Epidemiol
.
2017
;
31
(
5
):
452
462
26
Leung
DH
,
Heltshe
SL
,
Borowitz
D
, et al;
Baby Observational and Nutrition Study (BONUS) Investigators of the Cystic Fibrosis Foundation Therapeutics Development Network
.
Effects of diagnosis by newborn screening for cystic fibrosis on weight and length in the first year of life
.
JAMA Pediatr
.
2017
;
171
(
6
):
546
554
27
Mak
DY
,
Sykes
J
,
Stephenson
AL
,
Lands
LC
.
The benefits of newborn screening for cystic fibrosis: the Canadian experience
.
J Cyst Fibros
.
2016
;
15
(
3
):
302
308
28
Kasi
AS
,
Wee
CP
,
Keens
TG
,
Salinas
DB
.
Abnormal lung clearance index in cystic fibrosis screen positive, inconclusive diagnosis (CFSPID) children with otherwise normal FEV1
.
Lung
.
2020
;
198
(
1
):
163
167
29
Carlson
D
,
McKeen
E
,
Mitchell
M
, et al
.
Oropharyngeal flora in healthy infants: observations and implications for cystic fibrosis care
.
Pediatr Pulmonol
.
2009
;
44
(
5
):
497
502
30
Jourdain
S
,
Smeesters
PR
,
Denis
O
, et al
.
Differences in nasopharyngeal bacterial carriage in preschool children from different socio-economic origins
.
Clin Microbiol Infect
.
2011
;
17
(
6
):
907
914
31
Konno
M
,
Baba
S
,
Mikawa
H
, et al
.
Study of upper respiratory tract bacterial flora: first report. Variations in upper respiratory tract bacterial flora in patients with acute upper respiratory tract infection and healthy subjects and variations by subject age
.
J Infect Chemother
.
2006
;
12
(
2
):
83
96
32
Tumgor
G
,
Celik
U
,
Alabaz
D
, et al
.
Aetiological agents, interleukin-6, interleukin-8 and CRP concentrations in children with community- and hospital-acquired pneumonia
.
Ann Trop Paediatr
.
2006
;
26
(
4
):
285
291
33
Thauvin-Robinet
C
,
Munck
A
,
Huet
F
, et al;
Collaborating Working Group on R117H
.
The very low penetrance of cystic fibrosis for the R117H mutation: a reappraisal for genetic counselling and newborn screening
.
J Med Genet
.
2009
;
46
(
11
):
752
758
34
Anagnostopoulou
P
,
Latzin
P
,
Jensen
R
, et al
.
Normative data for multiple breath washout outcomes in school-aged Caucasian children
.
Eur Respir J
.
2020
;
55
(
4
):
55

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

Supplementary data