Page 174 - 2019_07 resto del Mondo-web

T. Dittrich et al.
In recent decades, biomarkers have gained essential importance for the diagnosis, risk stratification, response assessment and monitoring of AL patients. Besides the FLC test, which allows for highly sensitive quantification of the involved FLC and underlying clonal disorder,4–6 a wide range of organ biomarkers are available which reflect the severity of organ dysfunction and amyloid load in AL patients.7–14 Given the complexity of the disease, accurate estimates of prognosis are of great value to clinicians for optimal primary management and treatment.
Involvement of the heart and the consequent cardiac dysfunction are the major factors determining prognosis and, especially, early death of AL patients.15,16 Elevated car- diac biomarkers are, therefore, the most powerful prognos- tic determinants in AL. Serum levels of N-terminal pro-B- type natriuretic peptide (NT-proBNP) and cardiac troponin T (cTnT) were first found to predict survival in several cohorts of AL patients,10,11,15,16 and were consequentially incorporated into the first widely used staging system for AL, hereafter referred to as “MAYO2004”, which defined three stages.17
Subsequently, the composition and biomarker thresh- olds of the staging system for AL patients were refined.3,11,13,18–22 Two modifications have been widely accepted: the subgroup of patients with the most unfavor- able prognosis, who had elevated levels of both biomark- ers, could be further separated by an additional higher cut- off of serum NT-proBNP level (>8,500 ng/L) as well as a low systolic blood pressure.13 Patients classified as MAYO2004 stage III with NT-proBNP >8,500 ng/L had a dismal prognosis with a median overall survival of less than 5 months.13 The MAYO2004 system is, therefore, fre- quently applied with the modification that patients in stage III with an NT-proBNP >8,500 ng/L are assigned a “stage IIIb”, resulting in a system of four stages (hereafter referred to as “MAYO3b”).23 Concurrently, plasma cell fac- tors were incorporated by another four-stage system (“MAYO2012”) utilizing serum FLC level (threshold ≥180 mg/L) as well as NT-proBNP and troponins with different thresholds (NT-proBNP ≥1,800 pg/mL and cTnT ≥0.025 μg/L) compared to those of MAYO2004 (NT-proBNP ≥332 pg/mL and cTnT ≥0.035 μg/L).17,21
Given the importance of these three widely used staging systems for the management of AL patients, a systematic comparison of their performance is warranted. Moreover, serum levels of troponins and NT-proBNP are influenced by renal dysfunction and atrial arrhythmia or pacemaker stimulation, which is questioning the reliability of these biomarkers in subgroups with these concomitant disor- ders.24–28 Although impaired renal function and atrial arrhythmia are frequent in AL patients, neither of them was considered as a confounder during the development of the staging systems. We, therefore, systematically com- pared the performance of the MAYO2004, MAYO3b and MAYO2012 staging systems in a large cohort of newly diagnosed AL patients with special regard to the subgroups of patients with impaired renal function and atrial arrhyth- mia or pacemaker stimulation.
Methods
Patients and medical records
We identified 1,224 patients with confirmed AL, consecutively treated and followed at the Heidelberg Amyloidosis Center
between July 2002 and March 2017, for whom results of the required cardiac biomarkers and FLC assay were available before initial treatment. All demographic and clinical information, includ- ing age, gender, hematologic and organ-related laboratory test results, were obtained from medical records. Clinical data of half of the patients had not been used for statistical analysis before. This retrospective study was approved by the Ethics Committee of the University of Heidelberg and was conducted in accordance with the principles of the Helsinki declaration. All participants gave writ- ten informed consent to retrospective analysis of their clinical data.
Definitions
Organ involvement was defined according to consensus stan- dards.3 Detailed information about how the staging systems are applied and how atrial arrhythmia and impaired renal function are defined are provided in the Online Supplement.
Statistical analysis
Statistical analysis was performed using the R statistical envi- ronment,29 version 3.3.2 on a x86_64-w64-mingw32/x64 (64-bit) platform, together with the packages ‘survival’ (version 2.38), ‘VennDiagram’ (version 1.6.18) and ‘pec’ (version 2.5.4). Continuous data are described by their median and range. The Mann-Whitney U test was used to test differences in continuous variables and the χ2 test with continuity correction was used to test differences in categorical variables between two groups.
Overall survival curves were constructed according to Kaplan- Meier estimates and comparisons were made using the log-rank test. The median estimated time of observation was calculated based on the median time to censoring (reverse Kaplan-Meier). Univariable and multivariable Cox models were fitted to evaluate the influence of possible prognostic factors on overall survival. The proportional hazards assumption was tested as proposed by Grambsch & Therneau.30 To illustrate the results of the Cox mod- els, hazard ratios and corresponding 95% confidence intervals (95% CI) were calculated.
For the evaluation of the predictive accuracy of prognostic mod- els, a .632+ bootstrap estimate of the prediction error was calcu- lated for all event times using a time-dependent adaption of the Brier score.31 For the evaluation of discriminative accuracy, a .632 bootstrap estimate of the time-dependent adaption of the concor- dance index32 for time intervals of 6 months was calculated.
All statistical tests were two-sided. Results with P values not larger than 5% were considered statistically significant.
Further statistical methods are described in detail in the Online Supplement.
Laboratory methods
All biomarkers were measured by standard commercially avail- able assays. Reference values were as follows: FLCκ 3.3 – 19.4 mg/L, FLCλ 5.7 – 26.3 mg/L, NT-proBNP <125 ng/L and <334 ng/L (different assays), cTNT <0.03 μg/L, cardiac troponin I (cTNI) <0.04 μg/L and <0.6 μg/L (different assays), high sensitivity tro- ponin (hsTNT) <50 pg/mL. The cTNT assay was applied mainly until 2009, when it was replaced by assays with higher sensitivity or specificity, cTNI (since 2007) and hsTNT (since 2009). At least one troponin value was available for each included patient. The Modification of Diet in Renal Disease Study Group formula was applied to calculate the estimated glomerular filtration rate (eGFR).
Results
The baseline characteristics of the 1,224 patients includ- ed in the current study are reported in Table 1. The median age of the patients was 63 years and the majority (60%)
1452
haematologica | 2019; 104(7)