**21. Usefulness of clinical prediction tools in cystic fibrosis**

As previously outlined in this chapter, life expectancy continues to improve in patients with CF due to a combination of improvements in treatment, level of care, and diagnostic tools, as well as multidisciplinary input and the development of disease-specific centres. With contin‐ ually improving clinical outcomes, the need arises to better predict the prognosis of CF at an individual and group level. This need was first identified in 1958 with the development of the Shwachman-Kulczycki (SK) score to assess the severity of CF [139]. This score formed part of a study which monitored 105 patients for 5 years, the first of its kind to assess the long term progress of young CF patients. It was seen that there was increased frequency of CF patients surviving to adolescence and young adulthood and demonstrated a need to assess disease severity in order to predict the likely course of the disease.

There are many criticisms of the SK score as it is largely a subjective measure that depends on the clinical estimation of the examiner. Another drawback is that it does not include evalua‐ tions of pulmonary function. It was also devised with reference to a paediatric population in a time when life expectancy was drastically decreased. However, there have been a range of studies since its conception that have validated this score against a host of parameters increasingly used to measure prognosis. For example, Brasfield et al. observed a significant correlation between the SK score and chest radiography in an evaluation of over 640 chest radiogrpahs of 118 CF patients [130]. A more recent evaluation of the usefulness of this score

to assess disease severity was undertaken by Stollar et al. in 2011 where [140] significant correlations were demonstrated between FEV1, chest radiograph, HRCT, 6-minute walk test (6MWT), and SK score. It was felt that the SK score adequately reflects radiographic and functional impairments in patients with greater impairment of lung function, however, the score was less useful in patients with preserved lung function (FEV1 >70%). Regardless of its shortcomings, the SK score is regarded as a milestone in the history of CF and continues to serve as a somewhat tool in the determination of disease severity.

This score formed the basis for the development of a multitude of clinical prediction tools, the majority of which involve clinical parameters, radiological parameters, or a combination of both. One such early radiological scoring system is the Brasfield score [130], which is based on the chest radiograph findings of patients with CF and encompasses a maximum score of 25, with points scored according to the severity of air trapping, linear markings, nodular cystic lesions, large lesions, and an impression of the overall general severity of the radiograph. However, it similarly is a subjective scoring system and there can be inter-rater differences. The Brasfield score does correlate with pulmonary function tests and the SK score and is reproducible. A disadvantage of the Brasfield score is that it was developed to include radiographs assessed by a team of radiologists and clinicians, while the Northern score [141] is an advance on the Brasfield score which allows for a single examiner to determine the radiographic severity based on assessing chest radiographs divided into lung quadrants. The maximum score is 20, with higher scores reflecting poorer outcomes due to increased severity of disease. This score was validated against the Brasfield score and an earlier score, Chrispin-Norman [142], and was found to be equal in terms of consistency, reproducibility, and accuracy in reflecting overall clinical status as measured by the SK score.

The development of widespread use of computed tomography in clinical practice has led to the inclusion of HRCT in prognostic tools for CF. Nathanson et al. proposed the first such score which involved dividing the lung CT into 12 distinct zones in order to classify the severity of bronchiectasis on a 5-point scale in each zone [143]. This, once again, was validated against the SK and Brasfield scoring systems along with the results of pulmonary function tests carried out on the subjects. As discussed, Bhalla et al. [107] further developed the idea of CT scoring systems in order to aid selection for lung transplantation and put forward a score based on the severity of 9 different radiological parameters including peribronchial thickening, extent of bronchiectasis, and extent of mucus impaction. This score is very reproducible and correlates strongly with pulmonary function, hence is the gold standard for HRCT evaluation.

Clinically-based prognostic tools play a central role in the management of CF patients and serve to direct treatments, investigations, and multidisciplinary input. The National Institute of Health scale, developed in 1973, proposed a comprehensive scoring system which included multiple parameters including baseline demographics, various measures of lung function, and presence of common complications encountered in the disease [131]. A 100-point scale derived from these measurements correlated with severity of disease, the higher scores being associ‐ ated with poorer outcomes. It is a complex and cumbersome scale that reduces the likelihood for its application across the broader clinical setting. However, it is a reliable and reproducible score which encompasses many aspects of the disease and this has led to its continued use in CF-related research.

to assess disease severity was undertaken by Stollar et al. in 2011 where [140] significant correlations were demonstrated between FEV1, chest radiograph, HRCT, 6-minute walk test (6MWT), and SK score. It was felt that the SK score adequately reflects radiographic and functional impairments in patients with greater impairment of lung function, however, the score was less useful in patients with preserved lung function (FEV1 >70%). Regardless of its shortcomings, the SK score is regarded as a milestone in the history of CF and continues to

This score formed the basis for the development of a multitude of clinical prediction tools, the majority of which involve clinical parameters, radiological parameters, or a combination of both. One such early radiological scoring system is the Brasfield score [130], which is based on the chest radiograph findings of patients with CF and encompasses a maximum score of 25, with points scored according to the severity of air trapping, linear markings, nodular cystic lesions, large lesions, and an impression of the overall general severity of the radiograph. However, it similarly is a subjective scoring system and there can be inter-rater differences. The Brasfield score does correlate with pulmonary function tests and the SK score and is reproducible. A disadvantage of the Brasfield score is that it was developed to include radiographs assessed by a team of radiologists and clinicians, while the Northern score [141] is an advance on the Brasfield score which allows for a single examiner to determine the radiographic severity based on assessing chest radiographs divided into lung quadrants. The maximum score is 20, with higher scores reflecting poorer outcomes due to increased severity of disease. This score was validated against the Brasfield score and an earlier score, Chrispin-Norman [142], and was found to be equal in terms of consistency, reproducibility, and accuracy

The development of widespread use of computed tomography in clinical practice has led to the inclusion of HRCT in prognostic tools for CF. Nathanson et al. proposed the first such score which involved dividing the lung CT into 12 distinct zones in order to classify the severity of bronchiectasis on a 5-point scale in each zone [143]. This, once again, was validated against the SK and Brasfield scoring systems along with the results of pulmonary function tests carried out on the subjects. As discussed, Bhalla et al. [107] further developed the idea of CT scoring systems in order to aid selection for lung transplantation and put forward a score based on the severity of 9 different radiological parameters including peribronchial thickening, extent of bronchiectasis, and extent of mucus impaction. This score is very reproducible and correlates

strongly with pulmonary function, hence is the gold standard for HRCT evaluation.

Clinically-based prognostic tools play a central role in the management of CF patients and serve to direct treatments, investigations, and multidisciplinary input. The National Institute of Health scale, developed in 1973, proposed a comprehensive scoring system which included multiple parameters including baseline demographics, various measures of lung function, and presence of common complications encountered in the disease [131]. A 100-point scale derived from these measurements correlated with severity of disease, the higher scores being associ‐ ated with poorer outcomes. It is a complex and cumbersome scale that reduces the likelihood for its application across the broader clinical setting. However, it is a reliable and reproducible

serve as a somewhat tool in the determination of disease severity.

20 Cystic Fibrosis in the Light of New Research

in reflecting overall clinical status as measured by the SK score.

With the advancement of CF survival and life expectancy a need to develop simple and reliable prognostic tools for use in the clinical setting became apparent. In 1997, Hayllar et al. developed a predictive index based on research from over 400 patients and studied and correlated the index result with a mortality curve [144]. In this way, a simple calculation could be made to estimate 6-month and 1-year survival rates for a given score. The score was based on height, presence of hepatomegaly, white blood cell count, FEV1, and forced vital capacity (FVC). This, and scores like it, revolutionised the approach to prognosis in previously unchartered territory and survival estimates became central to prognostic tools. Liou et al. advanced on the NIH and Hayllar scores to present a 5-year prognostic tool, the longest range survival model in the field at its time of development [145]. It is based on a composite score of 8 clinical factors, as well as measurement of FEV1 and is a comprehensive, reproducible, reliable prognostic tool. However, it remains a complicated score to perform clinically and so focus has shifted in recent times to the development of simplified prediction tools that can be used within the time constraints of the current clinical setting. The CF-ABLE score was devised with this in mind and it involves the measurement of 4 of the major clinical parameters encountered in day to day practice [93]. Age, BMI, FEV1, and frequency of pulmonary exacerbations have been evaluated on a 7-point scale for correlation with prognosis and it has been found that patients with higher scores have a 26% chance of a poor outcome (death or transplantation) within 4 years of measurement. This score has been validated on a national registry and highlights the applicability of simple prediction tools that can be employed in the clinical environment.

Clinical and radiological prediction tools have emerged as a vital resource in the treatment of CF. They allow us to stratify patients by disease severity and to a certain extent allow us to predict likely adverse outcomes in patients with poor prognostic indices. A multitude of prognostic tools have been developed since the original SK score as CF research has continued to expand, each reflecting the changing face of CF treatment and diagnostic tools. Clinical practice favours the use of simple reliable scoring systems which take into account easily measurable parameters, such as the Hayllar score and the more up to date CF-ABLE score. Future research will need to involve the development of further prognostic tools to reflect the changes occurring in the management of CF (Table 2).




**Name of Score Year**

**Brasfield [130]**

**Nathanson [143]**

**Published**

22 Cystic Fibrosis in the Light of New Research

**NIH Score [131]** 1973 73 patients aged 3

1979 643 chest

1991 28 HRCT of CF patients

radiographs in 118 CF patients

**Number of patients in cohort**

to 30 years followed for a period between 3 and 6 years

**Parameters Used Strengths Limitations**

100 point scale of severity of lung and general parameters

as well as common complications: - Chest Radiograph - Pulmonary function - Pulmonary exacerbations




following: - Air trapping - Linear markings - Nodular cystic lesions - Large lesions - Overall impression of

severity


Composite score of presence and severity of each of the

Severity of bronchiectasis and mucous impaction on a 5 point scale as measured in 12 distinct lung zones










system - Overestimates rare clinical elements


**Table 2.** Summary of Clinical Prediction Tools in Cystic Fibrosis.
