|Home | About | Journals | Submit | Contact Us | Français|
A comparison of the longitudinal progression of lung disease in cystic fibrosis patients identified through newborn screening (NBS) in cohorts located in two different countries has never been performed and was the primary objective of this study.
The study included 56 patients in Brittany diagnosed through NBS between 1989 and 1994 and 69 similar patients in Wisconsin between 1985 and 1994. The onset and progression of lung disease was radiographically quantified using the Wisconsin Chest X-ray (WCXR) scoring system. A single pediatric pulmonologist blinded to all identifiers scored the films.
Generalized estimating equation analyses adjusted for age, genotype, sex, pancreatic insufficiency, and meconium ileus showed worse WCXR scores in Brittany patients compared to Wisconsin patients (average score difference=4.48; p<0.001). Percent predicted FEV1 was also worse among Brittany patients (p<0.001).
The finding of milder radiographically-quantified lung disease using the WCXR scoring system, as well as better FEV1 values, may be explained by variations in nutrition, environmental exposures, or healthcare delivery.
The regions of Wisconsin, United States, and Brittany, France have a very long history of newborn screening (NBS) for cystic fibrosis (CF), with screening having begun in 1985 in Wisconsin and in 1989 in Brittany (1-3). Each region also has extensive experience with longitudinal epidemiologic studies (3-9), and the two regions have collaborated on a genotype-pulmonary phenotype project (10). Although early diagnosis through NBS leads to uniformly better nutritional status (11) in all regions that have studied that outcome, there has been very little comparative evaluation of lung disease onset and progression after screening. To the best of our knowledge, a study comparing longitudinal pulmonary outcomes of children from regions of different countries has not yet been performed. Consequently, the regions of Wisconsin and Brittany afforded us an opportunity for such a study. Indeed, the comparability of the two populations seemed ideal to us for a study of longitudinal progression of pulmonary disease, the major source of morbidity and mortality in patients with CF (11), among patients from two regions of different countries.
Longitudinal quantitative chest radiography has been shown to be a sensitive method of detecting worsening pulmonary disease in children with CF, and correlates well with clinical variables (5, 12, 13). It is especially useful when study patients are too young to perform lung function tests, such as spirometry (13). Specifically, the Wisconsin Chest X-ray scoring system (WCXR) is more sensitive for detecting mild pulmonary disease in CF than the more commonly used Brasfield Chest X-ray scoring system (BCXR) (14, 15). We chose to use the WCXR in our study of longitudinal pulmonary outcomes.
Studies of CF that compare patients in different countries are rare. Our group performed a cross-sectional study of nutritional outcomes comparing US and Canadian CF patients by using data from the national registries in each country (16), which showed slightly better growth indices among Canadian patients compared to US patients. Another study was an international comparison of median age at death from cystic fibrosis (17). The authors concluded that among the 10 industrialized countries studied, median age at death was highest in the United States. However, France was not among the 10 countries because the data for France were in longer intervals than the other countries. Notably absent from the literature are longitudinal studies comparing CF patients who were found through newborn screening and closely followed over multiple years.
Because of the existence of specialized CF centres in both Brittany and Wisconsin, where the care is standardized (9, 18), and because of the similarity of the newborn screened populations, we hypothesized that the progression of lung disease using the WCXR scoring system would not differ significantly between CF patients in Brittany and Wisconsin.
The study included 56 CF patients who underwent NBS in Brittany between January 1989 and July 2000, and 69 Wisconsin patients who underwent NBS between April 1985 and July 1994. Both regions have complete NBS programmes covering their populations. In Brittany, NBS was carried out with immunoreactive trypsinogen (IRT) tests until September 1991, at which time IRT/DNA testing was instituted. Wisconsin also changed its NBS algorithm to IRT/DNA in July 1991 (4, 19), which ahs a sensitivity of about 95%. All screened patients had a confirmatory sweat test around 4–6 weeks of age, after which they began treatment for CF, since at the time of the study it was believed that younger infants would not be able to produce a sufficient volume of sweat for the test (20). All patients were followed at specialized CF centres in Wisconsin (Madison and Milwaukee) or Brittany (Roscoff, Rennes, Brest, Vannes, Lorient, Quimper, and St. Brieuc). Clinical management was similar between the two groups and included high energy intake, supplementation with pancreatic enzymes and fat-soluble vitamins, daily respiratory therapy, and treatment with antibiotics depending on respiratory symptoms and cultures; none of the centres used prophylactic anti-staph antibiotics. The staffing of the centres was also similar.
Lung disease was quantified using the WCXR scoring system described previously (14, 15) using original postero-anterior and lateral films. In Brittany, chest x-ray films were collected at diagnosis and then at ages 1, 2, 3, 4, 5, 6, 8, and 10–12 nearest the patient’s birthday when available. In Wisconsin, films were collected at the time of enrollment in the newborn screening study and then at ages 2, 4, 5, 6, 7, 8, 9, 10, 11, and 12 nearest the patient’s birthday. All films were scored by a single pediatric pulmonologist (PMF) blinded to all identifiers, including patient, centre, and country; this was accomplished by applying tape over all films and covering all identifiers. Wisconsin films were scored in Madison, while Brittany films were scored in Roscoff, France using a similar room, multi-panel viewing box, and a bright light. However, representative Wisconsin chest radiographs were deidentified and brought to Brittany for the scoring sessions there and then repeatedly mixed with those of French patients. The reproducibility of scoring was assessed by interspersing a subset of films that were scored previously from Wisconsin (n=10) and from Brittany patients (n=31). We have used this quality control strategy successfully on many occasions in the past (5, 7, 8, 14) and prefer it to other methods.
Data were collected on baseline risk factors, including sex, genotype, meconium ileus, pancreatic insufficiency, age at diagnosis, and weight and height at various ages. Genotype was determined as described elsewhere (10) and controlled for using three categories: 1) F508del homozygotes; 2) F508del compound heterozygotes; and 3) those with two other mutations (only 6% of the two cohorts). The molecular techniques used resulted in a very high CF mutation detection rate (99% in Brittany and >95% in Wisconsin) (21, 22). Weight-for-age and height-for-age percentiles were compared separately for male and female patients at ages 6 and 10 using French growth chart data for Brittany patients and U.S. growth chart data for Wisconsin patients. Data were also collected on time-dependent risk factors, including yearly FEV1 values and days of hospitalization per year. Percent predicted FEV1 values based on a patient’s age and height were calculated using international standards (23).
SAS version 9.10 (SAS Institute, Cary, NC) was used for all statistical analyses. Generalized estimating equation (GEE) models were used to compare longitudinal progression of WCXR scores and longitudinal progression of percent predicted FEV1 in the two cohorts. Another GEE model was constructed comparing longitudinal progression of WCXR scores among F508del homozygotes in the two populations. Additionally, a GEE model was constructed comparing WCXR scores among the larger CF centres (Madison, Milwaukee, Rennes, Roscoff, Brest, and St. Brieuc). Because of skewing in the data for number of days hospitalized between regular visits to CF centres, with many patients having no hospital days between visits, a GEE model comparing longitudinal odds of being hospitalized between visits was performed. Analyses were adjusted for sex, age, genotype, presence of pancreatic insufficiency, and presence of meconium ileus.
Characteristics of study patients are displayed in Table 1. There were no significant differences between Wisconsin and Brittany patients with respect to sex, genotype, presence of pancreatic insufficiency, presence of meconium ileus, or age at diagnosis. Using T-tests, male patients in Brittany did not differ significantly (p> .05) in weight or height from male patients in Wisconsin, although the country-specific percentile for the mean value of weight was somewhat lower for Wisconsin patients at age 10 (39th percentile) compared to Brittany patients at age 10 (49th percentile). Again using T-tests, female patients from Brittany were significantly shorter and weighed less than Wisconsin patients at both ages 6 and 10, and the country-specific percentiles for the mean values of weight and height were also lower for female Brittany patients than Wisconsin patients at ages 6 and 10. However, the lowest percentile for female patients in Brittany was for weight at age 10, which was the 37th percentile, showing that the average female Brittany patient in the study was not malnourished.
A total of 1003 chest radiographs were scored, an average of 8.88 radiographs per patient. Inclusion of 10 Wisconsin films that had already been scored in Wisconsin to the panel of films of Brittany patients yielded similar WCXR scores for the same film scored at each location (p=0.82). Likewise, inclusion of 31 Brittany films that had already been scored in Brittany to the panel of films yet to be scored resulted in similar WCXR scores for the same film scored two separate times (p=0.13).
All GEE models were restricted to age < 12.5, since no chest radiographs of patients older than 12 years of age were collected in Brittany. Longitudinal analysis of the progression of CXR scores is shown in Table 2. WCXR scores analyzed by region were significantly worse in Brittany compared to Wisconsin (p<0.001), with an average difference in score of 4.48. As expected, WCXR scores worsened with increasing age (p<0.001), with an average increase in WCXR score of 1.33 per year. Baseline risk factors including sex, genotype, the presence of pancreatic insufficiency, and the presence of meconium ileus were not significantly associated with WCXR scores. On the other hand, when either weight data or hospitalizations were added into the GEE model as potential explanatory variables, the effect of regional variations could not be detected. Therefore, the child’s weight (or hospitalization history) and region may have confounding effects; in other words the effect of region (Brittany or Wisconsin) on WCXR scores may be partly explained by weight and/or hospitalizations.
Longitudinal analysis of the progression of CXR scores among F508del homozygotes was performed separately. This analysis was intended to compare the progression of lung disease among subgroups of similar classic CF patients with a common genotype in Brittany and Wisconsin. Similar results were found with this analysis as were obtained with an analysis among all genotypes. More specifically, WCXR scores were significantly worse in Brittany compared to Wisconsin (p<0.0001), with an average difference in score of 5.95. As expected, WCXR scores worsened with increasing age (p<0.0001), with an average increase in WCXR score of 1.55 per year.
Longitudinal analysis of percent predicted FEV1 is shown in Table 3; the results are similar to those found with the WCXR scores. Brittany patients had worse percent predicted FEV1 values over time compared to Wisconsin patients (p<0.001); the average difference between the two populations was 11.56%. The correlation coefficient between percent predicted FEV1 and WCXR scores using 160 data points was −0.33 (p<0.0001). As expected, percent predicted FEV1 was worse with increasing age (p=0.05); the average difference per year was 0.85%. Patients with pancreatic insufficiency also had worse percent predicted FEV1 over time (p=0.05); the average difference between patients with pancreatic insufficiency and pancreatic sufficiency was 9.20%. Additionally, patients with meconium ileus had worse percent predicted FEV1 over time (p=0.043); the average difference between patients with meconium ileus and those without meconium ileus was 9.35%.
Table 4 shows percent predicted FEV1 values for Brittany and Wisconsin patients between ages 6 and 11. Percent predicted FEV1 is significantly lower for Brittany patients than for Wisconsin patients at all ages shown in the table.
GEE analysis of the longitudinal progression of WCXR scores by CF centre shows no significant differences among the 4 largest CF centres in Brittany. The other centres followed too few patients to be included in this analysis. When compared to the average WCXR score in Rennes, the difference in average score was −1.08 in Roscoff (p=0.47), 2.04 in St. Brieuc (p=0.26), and −4.70 in Vannes (p=0.44). There were significant differences between the average WCXR score in Rennes and the two Wisconsin CF centres, showing that the patients followed at the Wisconsin CF centres had significantly better WCXR scores than those followed in Rennes. When compared to the average WCXR in Rennes, the difference in average score was −4.70 in Madison (p=0.0006) and −4.00 in Milwaukee (p=0.0019).
Longitudinal analysis of the odds of hospitalization showed that Brittany patients were more likely than Wisconsin patients to be hospitalized between regular CF visits (p<0.0001; OR=18.82). Both a younger age and the presence of meconium ileus were associated with an increased likelihood of hospitalization between regular CF visits (for age: OR=0.95, p=0.016; for meconium ileus: OR=2.08, p=0.021). Sex, genotype, and presence of pancreatic insufficiency were not associated with the odds of hospitalization.
Figure 1 displays the mean WCXR scores at each age for Brittany and Wisconsin patients. In both populations, WCXR scores worsen with increasing age. WCXR scores are somewhat similar among the two populations until the age of 2, at which time Brittany patients show a more rapid worsening of scores. This increased rate of worsening occurs until age 6, at which time there is a stable worsening of scores in both populations.
Figure 2 displays the mean peribronchial thickening and bronchiectasis subcomponents of the WCXR scores at each age for Brittany and Wisconsin patients. In both populations, the peribronchial thickening score worsens with age, but among Brittany patients, there is a faster rate of development of peribronchial thickening compared to Wisconsin patients beginning at age 5. In both populations, the bronchiectasis score worsens with increasing age. However, there is a faster rate of development of bronchiectasis among the Brittany patients compared to the Wisconsin patients at all ages. The rapid rate of increase in bronchiectasis scores is most pronounced in patients under the age of 5.
To compare these CXR scores with results published by others, particularly the innovative strategy of McCormick et al (12) using centile charts, we transformed our WCXR scores into centile scores and then compared the two groups. For all the patients and all the periods up to age 12.5 years, the minimum is 0 and the maximum 52.95 in WCXR scores, and the mean WCXR score 13.81 for Brittany and 11.52 for Wisconsin patients. Using the minimum of 0 as the 0th centile and the maximum of 52.95 as 100th centile, we can transform all WCXR scores into centile scores, and then we calculate the 26.08th centile for the Brittany region and 21.75th centile for Wisconsin. The GEE model (adjusted for repeated measures) has an overall data difference of 4.48 or 4–6 score points, which corresponds to an overall mean difference of 13.81−11.52=2.29, which then corresponds to an overall centile difference of 26.08−21.75=4.33..
We believe that this international project is the first longitudinal study comparing pulmonary outcomes from CF patients diagnosed early through NBS in two different regions of two different countries. Unexpectedly, the results of our study clearly show that Wisconsin CF patients identified through newborn screening have a milder course of pulmonary disease than Brittany CF patients identified through newborn screening. Longitudinal chest x-rays scored with the WCXR scoring system demonstrate worse pulmonary disease among Brittany patients that develops within the first 6 years of life. The major subcomponent of the WCXR scoring system that determines this difference is bronchiectasis, as Brittany patients have a higher rate of development of bronchiectasis in their first 6 years of life than Wisconsin patients. Since bronchiectasis is an irreversible structural lung disease caused by persistent, severe infections (24), this is likely due to a higher rate of early infections among Brittany patients. Supporting our finding of milder pulmonary disease in Wisconsin patients compared to Brittany patients using WCXR scores are the results of percent predicted FEV1, which are better in Wisconsin patients, and lower odds of hospitalization in Wisconsin patients. One critique of the hospitalization results is likely baseline differential hospitalization rates in France, where the national health care system allows physicians to hospitalize patients without concern for medical costs footed by patients’ families, compared to the United States, where physicians may be more hesitant because of such medical costs. However, it is interesting to note that the improved condition of the Wisconsin patients is not attributable to increased hospitalizations. On the other hand, the confounding effect of hospitalization in the GEE model suggests that paradoxically such exposures associated with CF centre care may be detrimental.
Although we have clearly discovered milder pulmonary disease in Wisconsin compared to Brittany CF patients, we do not have an explanation for this difference from our current data but only hypotheses that can be studied in the future. Although we assumed that genotype could have been a potential reason for the difference, as suggested in several studies (24, 25), our sample size was limiting. Early nutritional deficiencies as shown by significantly subnormal weights or heights could have been another potential reason (27). Our GEE model incorporating the child’s weight with region and revealing a confounding effect supports this potential explanatory variable. However, the average weight and height of the patients at ages 6 and 10 were near the 50th country-specific percentile in both populations. Nevertheless, it is interesting to note that female Brittany patients have a lower average country-specific percentile for both weight and height than their Wisconsin counterparts, as well as the male Brittany patients.
Another hypothesis that may account for the difference is that environmental exposures leading to either increased infections or increased severity of infections with Pseudomonas aeruginosa and Staphylococcus aureus differ between the two populations. Unfortunately, the respiratory culture data available in this study were not comparable between the two populations, so we cannot state definitively that Brittany patients have more respiratory infections with these pathogens than Wisconsin patients. Several studies have shown that cross-infections can occur with social interactions and while at specialized CF centres (28-30). Although it is difficult, if not impossible, to quantify accurately the social interactions that may lead to cross-infection between CF patients retrospectively, all types of such social interactions are strongly discouraged at Wisconsin CF centres. On the other hand, at Brittany CF centres, patients are given guidelines to protect themselves when interacting with other CF patients but do not receive similar discouragement. In addition, they had more hospitalizations and perhaps potential for cross infection. In connection with this issue, Bush (31) has recently emphasize the hazards of CF centre care in the context of newborn screening and the concept of “first, do no harm.”
Thus, this study definitively demonstrated milder pulmonary disease among Wisconsin CF patients identified through NBS compared to a similar group of Brittany CF patients, but more research is needed to elucidate the cause of the findings. Moreover, since it is a retrospective study, it is unknown whether similar disparities and conditions exist today.
We would like to thank the following people for their help in this project: Centre de Référence Mucoviscidose, Nantes, France: Christine Grall and Laëtitia Guéganton; Inserm, U613, Brest, France: Claude Férec; Department of Nutritional Sciences, University of Wisconsin, Madison, Wisconsin: Hui-Chuan Lai. We would like to thank all the other CF practitioners (Dr. Anne Dirou, Jean Le Bihan, Sophie Ramel, Krista Revert), nurses (Marie Idres) and medical secretaries (Sophie Boubarne, Jocelyne Pengam) who contributed to data collection. We also would like to thank the French CF Registry for providing data (managed by the CF Association Vaincre La Mucoviscidose (Lydie Lemonnier) and operated by the National Demographic Institute (Gil Bellis)).
Funding Acknowledgement: This work was supported by National Institutes of Health grants DK 34108 and M01 RR03186, National Institutes of Health Clinical and Translational Science Award grant 1UL1RR025011, Cystic Fibrosis Foundation grant A001-5-01, the Association Vaincre La Mucoviscidose Clinical Research Hospital Program grant “Epidemiology of cystic fibrosis in western France, 2007”, and the University of Wisconsin Shapiro Summer Research Program grant.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflict of Interest Notification: None of the authors has a potential conflict of interest or financial relationship to disclose.