Validating quantitative PCR assays for cell-free DNA detection without DNA extraction: Exercise induced kinetics in systemic lupus erythematosus patients

Circulating cell-free DNA (cfDNA) has been investigated as a screening tool for many diseases. To avoid expensive and time-consuming DNA isolation, direct quantification PCR assays can be established. However, rigorous validation is required to provide reliable data in the clinical and non-clinical context. Considering International Organization for Standardization, as well as bioanalytical method validation guidelines we provide a comprehensive procedure to validate assays for cfDNA quantification from unpurified blood plasma. A 90 and 222 bp assay was validated to study the kinetics of cfDNA after exercise in patients with systemic lupus erythematosus. The assays showed ultra-low limit of quantification (LOQ) with 0.47 and 0.69 ng/ml, repeatability [≤] 11.6% (95% CI: 8.1-20.3), and intermediate precision [≤] 12.1% (95% CI: 9.2-17.7). Incurred sample reanalysis confirmed the precision of the procedure. The additional consideration of pre-analytical factors shows that centrifugation speed and temperature do not change cfDNA concentrations. In SLE patients cfDNA increases {approx}2 fold after all out walking exercise, normalizing after 60 min of rest. The established assays allow reliable and cost-efficient quantification of cfDNA in minute amounts of plasma in the clinical setting and can be used as a standard to control pre-analytical factors including cfDNA losses during purification.


28
Circulating cell-free DNA (cfDNA) is recognized to have reasonable prognostic or diagnostic potential in several 29 pathological diseases including cancer, sepsis, and autoimmune diseases such as SLE 1,2 . In 1966 Tan et al. 30 described high levels of cfDNA in serum of SLE patients 3 . Since that time, a plethora of studies confirmed elevated 31 cfDNA concentrations in serum and plasma of SLE patients with active or inactive disease (reviewed in 2 ). In 32 patients with autoimmune disease, high levels of cfDNA are considered to be a relevant antigen for auto anti-body 33 development 4-6 . Intriguingly, cfDNA concentrations increase about 10 fold during all out exercise, showing a half-34 life of ~15 min in the healthy population 7-9 . While regular physical exercise is recommended for SLE patients, and 35 several studies provided reasonable evidence that regular exercise reduces fatigue and increases cardiovascular 36 fitness in SLE patients 10 , the kinetics of cfDNA during exercise in SLE patients are still unknown. A minimal 37 invasive and cost-efficient assay is a valuable tool to monitor cfDNA concentrations, which might be indicative 38 for overtraining or disease remission 11,12 . 39 Most cfDNA quantification assays require pre-analytical DNA extraction 13 . A process, which is costly, time 40 consuming and has been shown to lead to variable loss of cfDNA depending on the extraction method 41 employed 14,15 . To avoid cfDNA isolation, direct quantification assays have been established 14,16 . Using multilocus 42 primers, which bind various sites in the human genome, a sufficient sensitivity can be reached. The hominoid 43 specific long interspersed element 1 (LINE1) family 2 (L1PA2) is a suitable target, according to its abundance and 44 specificity 17 . 45 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted January 20, 2021. ; https://doi.org/10.1101/2021.01.17.21249972 doi: medRxiv preprint NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.
A major prerequisite for reliable cfDNA detection is the evaluation of the assay performance. In 2019 the 46 International Organization for Standardisation (ISO) published a guideline which comprehensively specifies the 47 requirements for evaluating the performance of quantification methods for nucleic acid target sequences for 48 quantitative real-time PCR (qPCR) and digital PCR (dPCR) 18 . 49 Based on previous work 19 , we describe the validation of reliable qPCRs assays for the quantification of cfDNA. 50 Considering the relevant guidelines we emphasize the specificity, precision, including repeatability and 51 intermediate precision, limit of detection (LOD), limit of quantification (LOQ), linearity, as well as incurred 52 sample reanalysis. Since cfDNA typically shows a length of about 166bp 20 , the combination of a 90bp assay and 53 222bp assay enables the evaluation of DNA integrity 21,22 . 54 The assays could be applied successfully to quantify cfDNA samples in the clinical context. We show that exercise 55 induced cfDNA increases start to decline after 30 min in SLE patients, normalizing after 60 -90 min. Since 56 elevated cfDNA concentrations have been discussed to possibly trigger enhanced inflammation or the production 57 of anti-ds-DNA antibodies, low increases during and rapid decreases after exercise are rather positive aspects of 58 cfDNA kinetics during exercise in SLE patients. The direct quantification enables a time and cost saving 59 quantification assay, which further could be standard assay to control for pre-analytical factors including cfDNA 60 losses during purification. Furthermore, according to its high sensitivity the assay can be used to study cfDNA in 61 other body liquids, cell culture supernatant. 62 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

Methods
The copyright holder for this this version posted January 20, 2021. ; Blood sample collection. Venous blood samples from the cubital vein were collected before, directly after, 78 and 90 min after the exercise test. The utilized blood collection products are listed in Supplementary Table S1. 79 The samples were centrifuged immediately after drawing two times at 2500 x g for 15 min at room temperature 80 (RT), to collect platelet-free plasma. 24 Capillary blood samples were collected from the fingertip and stored at 4°C 81 before centrifugation at 600 x g for 10 min. For comparison of centrifugation protocols, venous blood samples 82 were centrifuged as indicated. 83 Assay validation material. Linearity and accuracy of the direct quantification method was tested on a custom-84 made 401-bp fragment from the L1PA2 family (Supplementary Table S2 In silico calculation of L1PA2 copy numbers per genome. To determine the number of targets for the 103 L1PA2_90bp and L1PA2_222bp primer pairs in the human genome (GRCh38/hg38), the UCSC Genome Browser 104 In-Silico PCR tool was used (https://genome.ucsc.edu/cgi-bin/hgPcr). The Max Product Size (Maximum size of 105 amplified region) was set to 660 bp, which equals the predicted maximal amplification rate of the polymerase 106 during 10s amplification. Min Perfect Match (Number of bases that match exactly on 3' end of primers) was set to 107 19 bp. The predicted sequences per chromosome and the length distribution are provided in Supplementary Fig.  108 S4 and Table S5. Determination of assay performance. For the determination of the dynamic range, assay linearity, LOD and 128 LOQ the purified L1PA2 DNA fragment was spiked into mouse plasma (1:10 final dilution). Mouse plasma is a 129 relevant sample matrix and does not contain the human specific L1PA2 sequences. Three independent standard 130 curves were prepared for each of the assays on different days. Each concentration (1 x 10 6 -25 copies per well) 131 was pipetted in septet replicates (Fig. 1). H 2 O and mouse plasma NTC controls, as well as the reference samples 132 were pipetted on each plate. For the determination of the LOQ the results from all standard curves were combined 133 to generate imprecision profiles using the Variance Function Program VFP (v1.2) with R (v3.6.3) (Fig. 1). The 134 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted January 20, 2021. y-intercept (Fig. 1). The efficiencies for the different standard curves are slightly lower than 90%. Ranging between 172 87.77 and 89.25% for the L1PA2_90bp assay and between 84.19 and 88.75% for L1PA2_222bp assay. 173 LOD and LOQ. All replicates of the low copy samples (25 copies) were detectable with the L1PA2_90bp and the 174 L1PA2_222bp assay, not reaching LOD (Fig. 1). Of note, as shown in Supplemental Fig. S6, the NTCs of the 175 L1PA2_90bp assay produce some background signal, which needs to be distinguishable from low copy samples. 176 Since the data were normally distributed with homogenous variances, the LOB and LOQ were calculated according 177 to the CLSI guideline 28 . The L1PA2_90bp assay shows a LOB of 8.39 copies and LOD of 18.59 copies. The LOQ, 178 which was derived from the imprecision profile was 32.5 copies for the L1PA2_90bp and 59.26 copies for the 179 L1PA2_222bp (Fig. 1), equaling a concentration of 0.47 ng/ml and 0.69 ng/ml, respectively. 180 Imprecision studies. The CVs were similar between the pooled or uniquely diluted plasma samples with 7.32% and 181 8.77% (PRE exercise sample), and 6.67% or 8.51% for the POST exercise sample (L1PA2_90bp assay) (Fig. 2a). 182 For the L1PA2_222bp the CVs were 9.17 and 9.37% for the PRE exercise and 7.55 and 8.56 for the post exercise 183 samples. The precision calculation for the measurements of the PRE and POST samples in repeated runs, resulted 184 in a repeatability ≤ 11.59% (95% CI: 8.10 -20.34) for the two assays. 185 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted January 20, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted January 20, 2021.  with the L1PA2_90bp assay after storage for > 2 month at -20°C. cfDNA concentrations did not differ significantly 217 (P = 0.35) and no degradation was obvious after a repeated freeze-thaw cycle. Prolonged storage of whole blood 218 (up to 180 min) before centrifugation did not lead to relevant cfDNA concentration changes ( Supplementary Fig.  219 S7). 220 Effects of cfDNA purification. The use of specific DNA isolation kits for cfDNA isolation from plasma 221 (Purification kit 1 and 2) slightly affects the cfDNA concentration and DNA integrity (Fig. 3a). During the isolation 222 process about 12% (± 9.5) and 23% (± 6.2) are lost with Purification kit 1 or 2, respectively. A kit for whole blood 223 DNA isolation (Purification kit 3) leads to relevant DNA losses of about 84% and increased DNA integrity, which 224 indicates a loss of short DNA fragments. 225 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted January 20, 2021. ; https://doi.org/10.1101/2021.01.17.21249972 doi: medRxiv preprint Influence of centrifugation speed or temperature. As shown in Fig. 3b and 3c, centrifugation speed and 226 temperature does not influence the cfDNA concentration in venous plasma. The concentration differences are 227 within the assay precision of the L1PA2_90bp assay. Notably, as shown in Fig. 3b in one subject the first of 4 228 taken samples shows higher cfDNA levels (independent from centrifugation speed), indicating that the first sample 229 should be discarded. 230 Incurred sample reanalysis. During the course of the study ~17% of the SLE study samples were reanalyzed. 231 The cfDNA concentrations correlate well between first and repeated analysis (r= 0.925, p<0.0001) (Fig. 2b). 232 Among the reanalyzed samples, 5 samples (9.16%) measured with the L1PA2_90bp assay show a difference > 233 ±30% between the initial and the repeated measurement (Fig. 2c). 25.53% of the reanalyzed samples show a 234 difference > ±30% between first and repeated measurement in the L1PA2_222bp assay. 14.1 ng/ml (95% CI: 10.5-19.9) 90 min after walking exercise (Fig. 4c). Similarly, the capillary samples increased 247 significantly ~ 2.2 fold from 11.1 ng/ml (95% CI: 8.0-15.4) to 24.8 ng/ml (95% CI: 17.9-34.4) and decreased to 248 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted January 20, 2021.  Fig. 4a and 4b display the 250 differences of cfDNA levels from the PRE value. 251 The comparison between venous and capillary cfDNA concentrations (Fig. 4d) shows an overall correlation of R 2 252 = 0.774, p < 0.0001 (dotted black line). The correlation between venous and capillary samples is higher for the 253 PRE exercise (R 2 = 0.933, p < 0.0001), and +90 min samples (R 2 = 0.974, p < 0.0001) compared to POST exercise 254 samples (ρ = 0.726, p = 0.0002). 255 As indicated in Fig. 5a

267
During the last years, cfDNA has been increasingly studied for its prognostic or diagnostic potential in various 268 clinical fields 2,32,33 . A prerequisite to use cfDNA levels for monitoring applications is to ensure reliable and 269 reproducible quantification 34  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted January 20, 2021. ; https://doi.org/10.1101/2021.01.17.21249972 doi: medRxiv preprint used representing low and high copy number samples. The CV for all samples, were below 10%, indicating 285 sufficient precision to clearly discriminate the effects of exercise. To determine the relative repeatability and 286 relative run-to-run variation standard deviations, duplicates of the low and high copy plasma pools were measured 287 in 10 consecutive qPCR runs. All precision estimations were below 12.1%. 288 Whenever samples are not analyzed in the same run, it is important to take inter-run calibration into account. For 289 qPCR assays this is of unique importance because the relationship between quantification cycle and relative 290 quantity is run dependent 37 . The baseline correction of each qPCR run is depending on the total number of samples 291 and the fluorescence intensity 37 . Here, we provided experimental proof that the use of two inter-run calibrators 292 reduced the intermediate precision between 8.07% and 16.29%. 293 A number of pre-analytical variables can affect the results of cfDNA measurement 15,38 . As reviewed by Ungerer 294 et al. 15 the influence of centrifugation speed is still discussed controversially, whereas a second centrifugation step 295 is recommended 38 . Our results indicate that centrifugation at 600 x g compared to 2500 or 16.000 x g does not 296 relevantly influence cfDNA concentrations. The detectable differences were within the assay precision. Notably, 297 we avoided untarred centrifuges, and disturbances of the cell pellet during pipetting. Similar to other studies we 298 did not detect cfDNA changes after prolonged storage of EDTA whole blood samples to several hours 39,40 . 299 According to the simpler, and less invasive sampling technique, capillary plasma samples are a reasonable sample 300 source, which can be collected more frequently. The comparison of venous and capillary samples shows high 301 congruence. Notably, repeatedly collected capillary samples show a higher variance of cfDNA concentrations 302 compared to venous samples, which is not related to assay imprecision (as indicated in Supplementary Fig. S7). 303 The reason is not clarified in detail, and the intra-individual variance needs further systematic evaluation. 304 Studies indicate that SLE patients show higher cfDNA concentrations compared to the healthy population, and 305 that cfDNA concentrations are related to disease activity 12,41 . Exercise is recommended for the treatment of SLE 306 patients 42 . However, the kinetics of cfDNA after exercise has not been studied. After walking until exhaustion, the 307 studied SLE patients showed lower fold-changes compared to healthy subjects, which is likely related to exercise 308 intensity, time until exhaustion, and total energy expenditure 7,14,43 . cfDNA concentrations normalized 60-90 min 309 after exercise in most of the patients (Fig. 4 and Fig. 5), showing no significant differences from the PRE cfDNA 310 values (Supplementary Tables S8 and S8 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted January 20, 2021. ; https://doi.org/10.1101/2021.01.17.21249972 doi: medRxiv preprint