Predictive factors for treatment response and mortality in chronic pulmonary aspergillosis

Chronic pulmonary aspergillosis (CPA) is associated with significant mortality, and suboptimal antifungal treatment response. We describe predictive factors for treatment response and survival.


| INTRODUC TI ON
Chronic pulmonary aspergillosis (CPA) is an indolent fungal infection which usually affects patients with underlying structural lung disease who are not overtly immunocompromised.3][4][5] Most studies assessing treatment response have used a combination of clinical and radiological parameters although none of these characteristics, either individually or in combination, have been shown to reliably predict prognosis.
Previous studies have described radiological changes in cohorts of CPA patients following antifungal therapy and have attempted to correlate them with clinical outcomes.Godet et al. 6 reported that clinical response was associated with loss of fungal ball and reduction in cavity wall and pleural thickness.In their cohort of CPA patients, Nam et al. 7 found that while antifungal therapy resulted in decreased cavity wall thickness, this was not associated with clinical response, while initial large cavity size was associated with clinical deterioration.These studies used assessments of clinical response based in part on patient reported clinical symptoms, which may be affected by progression of underlying lung disease, for example COPD or sarcoidosis, independent from the progression or otherwise of CPA.
Relatively high mortality rates have been reported in CPA patients.Lowes et al found a mortality of 14% at 1 year and 38% at 5 years, while Nakamoto et al. reported a mortality of 52% at 5 years. 8,9However, it is not known to what extent the main driver of this mortality is the chronic fungal infection itself or the underlying predisposing disease.For instance, Lowes et al. 9 identified COPD as a significant risk factor for mortality.If CPA contributes to mortality, then early introduction of antifungal therapy would be expected to have a significant effect on survival.Here we describe predictive factors for treatment response and mortality in a cohort of patients with CPA who had serial chest CT imaging prior to and following antifungal therapy.We hypothesize that certain CPA-specific or underlying disease-specific factors may affect response to treatment.
Increased understanding of any such predictive factors would enable a more personalised approach to treatment, such as more intensive antifungal therapy in those with poor prognostic factors.Unlike previous similar studies we assess Aspergillus serology as a potential predictive factor.Respiratory Society (ESCMID/ERS) CPA guidelines. 1 Patients were included for analysis if they had a CT chest scan done as part of routine care within 6 months prior to starting oral antifungal therapy and a second scan for assessment of response performed between 3 and 12 months after commencing antifungal therapy.Inclusion criteria also included a minimum of 2 months cumulative antifungal therapy between first and second scans.Subjects also had to have had no antifungal therapy for at least 6 months prior to the first scan.

| ME THODS
Radiological assessment for presence or absence of fungal balls, maximum diameter of any fungal balls, maximal pleural thickness, cavity diameter and cavity wall thickness were all based on contemporaneous reports made by Consultant Radiologists.The overall impression of the reporting radiologist regarding improvement, stability or deterioration of CPA was used.If no overall impression was given by the radiologist, the scans were classified as improved if a decrease in one of the following parameters occurred post-treatment: maximal cavity wall thickness, maximal pleural thickness, aspergilloma size or maximal cavity diameter and no deterioration in one of these parameters.The scan was classified as stable if all the above parameters were reported as stable.The scan was classified as deteriorating if at least one of the parameters was reported as deteriorating (Figures S1   and S2).Any reduction in the parameters was classified as improvement and any increase was classified as deterioration.
Data were collected for patient demographics, underlying diseases, Charlson comorbidity index, antifungal therapy, Aspergillus serology (Aspergillus IgG, Aspergillus IgE and total IgE), inflammatory markers, albumin, weight, Medical Research Council (MRC) dyspnoea score where available.Aspergillus serology was performed with the ImmunoCAP assay.Aspergillus IgG was reported as >200 mg/L in five patients (in all five it was the first measurement, and in one of them also the repeat measurement after treatment).
In those cases, a value of 200 mg/L was accepted for the analysis.

Serological improvement was taken as at least a 50% reduction in
Aspergillus IgG or a return to a normal level (<40 mg/L) in accordance with the CPAnet consensus statement. 2In addition to the MRC dyspnoea score, the clinician's overall assessment of clinical response at the 6-month visit was recorded.Clinical improvement was taken as an improvement in at least one symptom felt to be related to CPA, without deterioration in another symptom or documentation of the clinician's impression of clinical improvement in the notes.Clinical deterioration was taken as a deterioration in at least one CPA-related symptom, or documentation of the clinician's impression of clinical deterioration in the notes.Combining the clinical and radiological parameters, 'favourable response' was defined as improved/stable clinical status and improved/stable radiological status, whereas 'unfavourable response' was defined as deterioration in clinical or radiological status.This study was a retrospective service evaluation and ethical approval was not required.The primary outcome were the parameters associated with a favourable response and mortality.Multivariate analysis of factors contributing to mortality included the parameters found to be significant to the level of p < .05 on univariate analysis.Survival analyses were performed using Cox regression.For analysis of survival according to patient baseline factors, follow up time was defined as the time from initiation of antifungal therapy to time of death or last known follow up point if patient was alive.For analysis of survival according to parameters of response to antifungal therapy, follow up time was defined as the time from the CT scan done to assess the effect of therapy to the time of death or last known follow up point if patient was alive.For the purpose of survival analysis, Aspergillus IgG and C-reactive protein (CRP) values were log-transformed to base-10 as they were skewed (there were eight patients with very high Aspergillus IgG, >500 mg/L and four patients with very high CRP, >125 mg/L).

| Clinical outcomes
On clinical assessment at 6 months of antifungal therapy, 21 (36%) had clinically improved, 20 (34%) were clinically stable and 15 (25%) had deteriorated, while the clinical outcome was not recorded for three (5%) cases.Among those who improved, the clinical impression of improvement was due to reported improvement in weight in seven cases, cough in five, sputum in four, haemoptysis in three and dyspnoea in one.In six cases, the documentation was that of a general improvement, without mention of improvement in specific symptoms.
Among those who deteriorated, the clinical impression of deterioration was due to reported worsening of the following symptoms: ongoing weight loss in eight cases, haemoptysis in five, dyspnoea in four, fatigue in two and cough in one.Table 1 shows relevant clinical and laboratory parameters at baseline and at the 6-month assessment.
Aspergillus IgG declined following antifungal therapy in 89% of cases for which paired measurements were available.Only 32% of cases had serological improvement as defined by CPAnet consensus.

| Radiological outcomes
Antifungal treatment was started a mean of 1.9 months (range 0-6) after a baseline chest CT scan was obtained, with the follow up chest CT scan performed a mean of 7.3 months (range, 3-12) after starting treatment.

| Overall outcomes
The overall response could be estimated in 55 patients.Of these, 34 (62%) had a favourable response.The association of overall (radiological and clinical) response with baseline parameters is shown in TA B L E 1 Baseline characteristics of patients with CPA.

| Survival
Mean follow up time was 35.6 (SD 15.5) months and 7% were dead after 1 year.In total 24 (41%) patients died during follow up.The cause of death could not be confirmed as most patients died in other facilities or in the community.Table 2 shows a univariate analysis of the effect of patient baseline parameters on survival.
Parameters with a p < .05were included in multivariate analysis (Table 3).
Figure 1 shows survival according to the overall response status.

| DISCUSS ION
Assessment of treatment response in CPA is challenging, as patients often have multiple comorbidities, severe underlying lung disease and are prone to recurrent bacterial and viral chest infections. 10In TA B L E 2 Clinical and radiological parameters at baseline as predictors of favourable response and survival following initiation of antifungal therapy.

Response assessment Survival
Overall routine care, a combination of clinical, radiological and serological criteria may be used.We used a composite end point of clinical and radiological response based on data collected during routine care.
We found this end point to be predictive of mortality, suggesting it is clinically relevant and can be used in routine care to assess treatment response in CPA.We used clinical improvement/stability and radiological improvement/stability as our definition of a favourable outcome, similar to the criteria used by Sehgal et al. 11 In most CPA studies, response rates are around 60%, suggesting suboptimal responses with all antifungal agents, therefore having predictors of treatment response would be desirable.We found a higher CRP to be the only significant predictor or unfavourable response, although we observed a trend for patients with a total IgE of >300 IU/mL and presence of Aspergillus sensitisation to have a favourable response.Aspergillus sensitisation is common in but its significance is not clear. 12Although an exaggerated Th2 response is known to be the main feature in ABPA, it may also be contributing to CPA pathogenesis: a histological study of resected aspergillomas showed the presence of IgE in tissue with a possible role in mast cell activation in patients with a high total IgE. 13Sehgal et al. 12 did not observe a difference in outcomes at 6 months between patients with high or low total IgE, although they did observe significantly longer time to relapse in those with high IgE.
Clinical improvement is difficult to assess in CPA.Using the clinician's impression of clinical improvement, only 36% improved.
This may reflect the large number of patients with underlying COPD whose baseline symptoms may have not been affected by antifungals.However, we observed that clinical improvement provided a survival advantage compared to stability (Figure 2).Overall treatment response, as defined by CPAnet consensus, includes clinical improvement (rather than improvement OR stability), which therefore may be a more relevant prognostic indicator. 2We could not apply the CPAnet consensus criterion in this study, as this would require a structured assessment of patients' symptoms which was not available in this retrospective study. 14e most objective assessment of treatment response is radiology.We found enlarging cavity size, enlarging aspergilloma and progressing cavity wall thickness on treatment to all be linked with mortality.The association was strongest with enlarging cavity size.
This finding is consistent with the observation by Nam et al. 7  F I G U R E 1 Kaplan-Meier curve of survival according to overall response (favourable response defined as: clinically better/stable and radiologically better/stable).predictor of mortality (Figure 2B); therefore, maintaining radiological stability may be a desirable outcome after a trial of antifungals, particularly in cases where deterioration was occurring before antifungals were started.
Serological response (reduction in Aspergillus IgG by 50% or return to normal levels, as suggested by CPAnet consensus) did not correlate with clinical/radiological response. 2This validates the CPAnet recommendation not to consider serology to define overall treatment response.Interestingly, there was a trend for higher mortality in patients who had serological response; this may reflect a higher baseline Aspergillus IgG (Figure 2C).Indeed, we found Aspergillus IgG an adverse prognostic factor for mortality.inhaled steroids may lead to boosting of steroid levels to varying degrees.This would be in contrast to the accepted notion that steroids are detrimental in aspergillosis.In addition, there is some evidence that inhaled corticosteroids reduce mortality and exacerbation frequency in COPD, a co-morbidity that affected just over half of patients in this cohort. 15,16is study is limited by the relatively small number of patients, since we included only patients who had CT scans performed within a specified time frame.In routine care, scans may not be performed as rigorously as during a prospective study.For this reason, the radiological outcomes should be interpreted with caution.We relied on the interpretation of the radiologist reporting the scans during routine clinical care and not specifically for the purposes of the study.This meant that certain features of CPA, such as pleural thickening, or cavity volume, were not specifically reported and we could not estimate their contribution to prognosis.However, it is likely that the radiological assessments used in this study reflect real-world outcomes assigned to patients in routine clinical practice and which go on to affect subsequent treatment decisions.In addition, clinical assessment of response was based on physicians' impressions during routine clinical care and not on a standardised questionnaire such as the respiratory symptom score recommended by CPAnet. 2 This would have been possible in a prospective study.In our patient population, COPD was the most common condition predisposing to CPA; this does not apply to many other settings where prior TB is more common and the patient population may be younger and with fewer comorbidities.Finally, this retrospective study was not designed to compare different antifungal treatments and we did not analyse therapeutic drug level monitoring.In our clinic, voriconazole was prescribed in patients with lower body weight, higher Aspergillus IgG, higher CRP and lower albumin, suggesting that voriconazole was preferred by the clinician for those perceived as having more severe disease.
In conclusion, a favourable clinical and radiological response after antifungal treatment was associated with reduced mortality in CPA.Mortality was higher in those with higher inflammatory markers, lower albumin and higher Aspergillus IgG, whereas we found no correlation with underlying lung disease such as COPD.This suggests that mortality is driven mostly by CPA rather than the underlying disease and that early optimal CPA treatment may improve outcomes.

Factors
predicting overall response and mortality were assessed using SPSS version 2 for Windows.Chi-square was used for categorical variables, t-test for continuous parametric variables and the Mann Whitney U or McNemar test for non-parametric variables.
Kaplan-Meier curve of survival according to clinical response at 6 months.(B) Kaplan-Meier curve of survival according to radiological response at 6 months.(C) Kaplan-Meier curve of survival according to serological response at 6 months.
3 (A) Kaplan-Meier curve of survival according to change in cavity wall thickness.(B) Kaplan Meier curve of survival according to change in cavity size.(C) Kaplan-Meier curve of survival according to change in aspergilloma size.
Patients who were commenced on oral triazole antifungal therapy for CPA between May 2018 and April 2021 were identified from the National Aspergillosis Centre (Manchester, UK) patient database.The diagnosis of CPA was made in accordance with the European Society of Clinical Microbiology and Infectious Diseases and European

Table 2 .
Serological response as defined by CPAnet criteria did not correlate with overall outcomes: among those with favourable Note: Data available for 59 patients unless otherwise specified in brackets in the first column.Numbers are n (%) or mean ± standard deviation unless otherwise specified.Abbreviations: IQR, interquartile range; MRC, Medical Research Council; NTM, non-tuberculous mycobacterial.
Multivariate analysis of factors for survival.