Impact of Changes of the 2020 Consensus Definitions of Invasive Aspergillosis on Clinical Trial Design: Unintended Consequences for Prevention Trials?

BACKGROUND: Consensus definitions for the diagnosis of invasive fungal diseases (IFDs) were updated in 2020 to increase the certainty of IFD for inclusion in clinical trials, for instance by increasing biomarker cutoff limits to define positivity. To date, there is a paucity of data as to the impact of the revised definitions on clinical trials. METHODS: In this study, we sought to determine the impact of the new definitions on classifying invasive aspergillosis (IA), the most common invasive mold disease in immunocompromised patients. We reclassified 226 proven and probable IA cases plus 139 possible IFD cases in the Aspergillus Technology Consortium (AsTeC) and in an antifungal prophylaxis trial (BMT CTN 0101) using the new criteria. RESULTS: Fewer cases met the more stringent diagnostic 2020 criteria after applying the reclassification criteria to define probable IA. Of 188 evaluable probable cases, 41 (22%) were reclassified to 40 possible IA and 1 probable IFD. Reclassification to possible IFD occurred in 22% of hematologic malignancy (HM) patients, 29% of hematopoietic cell transplant (HCT) patients, and in no lung transplant (LT) patients. Date of diagnosis was established a median (range) of 3 (1-105) days later in 15% of probable IA cases using the new criteria. Applying the new definitions to the BMT CTN 0101 trial, the power to detect the same odds ratio decreased substantially. CONCLUSIONS: The updated IA consensus definitions may impact future trial designs, especially for antifungal prophylaxis studies.

Severity of Acute Graft-versus-Host Disease and Associated Healthcare Resource Utilization, Cost, and Outcomes.

Acute graft-versus-host disease (aGVHD) contributes to poor outcomes and increased healthcare resource utilization (HRU) after allogeneic hematopoietic stem cell transplantation (HCT). However, HRU and the economic burden of aGVHD based on severity of the disease is not well characterized. Our study cohort comprised 290 adults who underwent allogeneic HCT between 2010 and 2018. Costs, HRU, and all-cause mortality in the 100-day and 365-day periods after HCT were compared between patients with aGVHD and those without aGVHD. The impact of aGVHD severity and gastrointestinal (GI) involvement on mortality, HRU, and economic burden was also evaluated. Medical costs and total hospital length of stay (LOS) were retrieved from administrative data that allocate costs to services based on departmental input for resource use and were adjusted to 2018 dollars. The Wilcoxon rank-sum test was used to compare the number of inpatient days and total costs. Multivariable linear regression was fitted on log-transformed costs. Compared with patients without aGVHD, those with aGVHD had a significantly greater median hospital LOS (28 days versus 22 days) and higher rates of intensive care unit (ICU) admission (13% versus 6%) and rehospitalization (59% versus 38%) during the first 100 days post-HCT. The presence of grade I-II aGVHD significantly prolonged the hospital LOS by a median of 3 days and increased the readmission rate by 18%, whereas grade III-IV aGVHD was associated with a nearly 30% increase in the readmission rate and a doubling of inpatient LOS, ICU admission rate, and mortality in the first 100 days post-HCT. Compared with the absence of aGVHD, lower GI involvement in aGVHD was also associated with increased risk of readmission (30%) and twice as many inpatient days, doubling the likelihood of ICU admission and mortality over the first 100 days. Similar findings were observed over days 101 to 365 post-HCT. The mean cost attributable to aGVHD regardless of grade was $60,923 in the first 100 days post-HCT. This cost varied by grade. The mean aGVHD- attributable costs were $18,071 for grade I, $36,115 for grade II and $120,929 for grade III/IV aGVHD and $114,668 for aGVHD involving the lower GI tract. In the 101- to 365-day period, the mean attributable aGVHD cost regardless of grade was $17,527. This cost also varied by grade. There were no additional aGVHD-attributable costs for grade I, but the mean aGVHD-attributable costs were $9743 for grade II, $62,220 for grade III/IV, and $55,724 for aGVHD with lower GI involvement compared with the controls without aGVHD. High-grade aGVHD and GI involvement in aGVHD, especially lower GI aGVHD, is associated with a considerably increased mortality and healthcare economic burden. Therefore, it is imperative that new therapeutic strategies be developed for this patient population.

A genomics-informed computational biology platform prospectively predicts treatment responses in AML and MDS patients.

Patients with myelodysplastic syndromes (MDS) or acute myeloid leukemia (AML) are generally older and have more comorbidities. Therefore, identifying personalized treatment options for each patient early and accurately is essential. To address this, we developed a computational biology modeling (CBM) and digital drug simulation platform that relies on somatic gene mutations and gene CNVs found in malignant cells of individual patients. Drug treatment simulations based on unique patient-specific disease networks were used to generate treatment predictions. To evaluate the accuracy of the genomics-informed computational platform, we conducted a pilot prospective clinical study (NCT02435550) enrolling confirmed MDS and AML patients. Blinded to the empirically prescribed treatment regimen for each patient, genomic data from 50 evaluable patients were analyzed by CBM to predict patient-specific treatment responses. CBM accurately predicted treatment responses in 55 of 61 (90%) simulations, with 33 of 61 true positives, 22 of 61 true negatives, 3 of 61 false positives, and 3 of 61 false negatives, resulting in a sensitivity of 94%, a specificity of 88%, and an accuracy of 90%. Laboratory validation further confirmed the accuracy of CBM-predicted activated protein networks in 17 of 19 (89%) samples from 11 patients. Somatic mutations in the TET2, IDH1/2, ASXL1, and EZH2 genes were discovered to be highly informative of MDS response to hypomethylating agents. In sum, analyses of patient cancer genomics using the CBM platform can be used to predict precision treatment responses in MDS and AML patients.

Applying host disease status biomarkers to therapeutic response monitoring in invasive aspergillosis patients.

One critical factor impeding successful management of invasive aspergillosis (IA) is the lack of reliable biomarkers to assess therapeutic response. We hypothesized that changes in certain host biomarkers reflect the nature of infection status and disease progression. Upon primary IA diagnosis, these disease status biomarkers can be monitored to track response to antifungal therapy and provide early markers that prognosticate likelihood of response. Herein, we analyzed serum levels of three prominent host disease status biomarkers C-reactive protein (CRP), haptoglobin (Hp), and annexin A1 (ANXA1) in IA patients during antifungal therapy. A total of 81 serial serum samples were collected at five or six different time points relative to IA diagnosis from 15 probable IA patients (10 acute leukemia [AL] and five hematopoietic stem cell transplantation [HSCT]). Of note, different biomarker profiles were observed in AL and HSCT patients, as not only levels of markers were significantly lower in HSCT patients but also more prominent interconnections among markers were observed in AL patients. Using a composite evaluation, patients were categorized as responders, nonresponders, and stable cases at last specimen. For AL responders, typical biomarker profiles were high initially but rapidly decreased for CRP and Hp post antifungal therapy, while low initial ANXA1 values were restored to normal levels after treatment. In contrast, CRP and Hp were persistently elevated whilst ANXA1 remained low throughout therapy in AL non-responders. As a pilot proof-of-concept study, our work demonstrates the great potential of using host biomarkers to monitor early therapeutic response in leukemia patients.

Characteristics of Late Fatal Infections after Allogeneic Hematopoietic Cell Transplantation.

We analyzed late fatal infections (LFIs) in allogeneic stem cell transplantation (HCT) recipients reported to the Center for International Blood and Marrow Transplant Research. We analyzed the incidence, infection types, and risk factors contributing to LFI in 10,336 adult and 5088 pediatric subjects surviving for ≥2 years after first HCT without relapse. Among 2245 adult and 377 pediatric patients who died, infections were a primary or contributory cause of death in 687 (31%) and 110 (29%), respectively. At 12 years post-HCT, the cumulative incidence of LFIs was 6.4% (95% confidence interval [CI], 5.8% to 7.0%) in adults, compared with 1.8% (95% CI, 1.4% to 2.3%) in pediatric subjects; P < .001). In adults, the 2 most significant risks for developing LFI were increasing age (20 to 39, 40 to 54, and ≥55 years versus 18 to 19 years) with hazard ratios (HRs) of 3.12 (95% CI, 1.33 to 7.32), 3.86 (95% CI, 1.66 to 8.95), and 5.49 (95% CI, 2.32 to 12.99) and a history of chronic graft-versus-host disease GVHD (cGVHD) with ongoing immunosuppression at 2 years post-HCT compared with no history of GVHD with (HR, 3.87; 95% CI, 2.59 to 5.78). In pediatric subjects, the 3 most significant risks for developing LFI were a history of cGVHD with ongoing immunosuppression (HR, 9.49; 95% CI, 4.39 to 20.51) or without ongoing immunosuppression (HR, 2.7; 95% CI, 1.05 to 7.43) at 2 years post-HCT compared with no history of GVHD, diagnosis of inherited abnormalities of erythrocyte function compared with diagnosis of acute myelogenous leukemia (HR, 2.30; 95% CI, 1.19 to 4.42), and age >10 years (HR, 1.92; 95% CI, 1.15 to 3.2). This study emphasizes the importance of continued vigilance for late infections after HCT and institution of support strategies aimed at decreasing the risk of cGVHD.

PCR-Based Approach Targeting Mucorales-Specific Gene Family for Diagnosis of Mucormycosis.

Mucormycosis is an aggressive, life-threatening infection caused by fungi in the order Mucorales. The current diagnosis of mucormycosis relies on mycological cultures, radiology and histopathology. These methods lack sensitivity and are most definitive later in the course of infection, resulting in the prevention of timely intervention. PCR-based approaches have shown promising potential in rapidly diagnosing mucormycosis. The spore coating protein homolog encoding CotH genes are uniquely and universally present among Mucorales. Thus, CotH genes are potential targets for the rapid diagnosis of mucormycosis. We infected mice with different Mucorales known to cause human mucormycosis and investigated whether CotH could be PCR amplified from biological fluids. Uninfected mice and those with aspergillosis were used to determine the specificity of the assay. CotH was detected as early as 24 h postinfection in plasma, urine, and bronchoalveolar lavage (BAL) samples from mice infected intratracheally with Rhizopus delemar, Rhizopus oryzae, Mucor circinelloides, Lichtheimia corymbifera, or Cunninghamella bertholletiae but not from samples taken from uninfected mice or mice infected with Aspergillus fumigatus Detection of CotH from urine samples was more reliable than from plasma or BAL fluid. Using the receiver operating characteristic method, the sensitivity and the specificity of the assay were found to be 90 and 100%, respectively. Finally, CotH was PCR amplified from urine samples of patients with proven mucormycosis. Thus, PCR amplification of CotH is a promising target for the development of a reliable, sensitive, and simple method of early diagnosis of mucormycosis.

Conditioning chemotherapy dose adjustment in obese patients: a review and position statement by the American Society for Blood and Marrow Transplantation practice guideline committee.

Hematopoietic stem cell transplantation (HCT) is a potentially life-saving therapy for patients with malignant and nonmalignant disease states. This article reviews the current published literature on the dosing of pharmacologic agents used for HCT preparative regimens with specific focus on the obese patient population. The review found that dose adjustments for obesity have, to date, been based empirically or extrapolated from published data in the nontransplantation patient population. As a result, the Committee determined that clear standards or dosing guidelines are unable to be made for the obese population because Level I and II evidence are unavailable at this time. Instead, the Committee provides a current published literature review to serve as a platform for conditioning agent dose selection in the setting of obesity. A necessary goal should be to encourage future prospective trials in this patient population because further information is needed to enhance our knowledge of the pharmacokinetics and pharmacodynamics of conditioning agents in the setting of obesity.

Randomized, double-blind trial of fluconazole versus voriconazole for prevention of invasive fungal infection after allogeneic hematopoietic cell transplantation.

Invasive fungal infection (IFI) is a serious threat after allogeneic hematopoietic cell transplant (HCT). This multicenter, randomized, double-blind trial compared fluconazole (N = 295) versus voriconazole (N = 305) for the prevention of IFI in the context of a structured fungal screening program. Patients undergoing myeloablative allogeneic HCT were randomized before HCT to receive study drugs for 100 days, or for 180 days in higher-risk patients. Serum galactomannan was assayed twice weekly for 60 days, then at least weekly until day 100. Positive galactomannan or suggestive signs triggered mandatory evaluation for IFI. The primary endpoint was freedom from IFI or death (fungal-free survival; FFS) at 180 days. Despite trends to fewer IFIs (7.3% vs 11.2%; P = .12), Aspergillus infections (9 vs 17; P = .09), and less frequent empiric antifungal therapy (24.1% vs 30.2%, P = .11) with voriconazole, FFS rates (75% vs 78%; P = .49) at 180 days were similar with fluconazole and voriconazole, respectively. Relapse-free and overall survival and the incidence of severe adverse events were also similar. This study demonstrates that in the context of intensive monitoring and structured empiric antifungal therapy, 6-month FFS and overall survival did not differ in allogeneic HCT recipients given prophylactic fluconazole or voriconazole. This trial was registered at www.clinicaltrials.gov as NCT00075803.

Use of bronchoalveolar lavage to detect galactomannan for diagnosis of pulmonary aspergillosis among nonimmunocompromised hosts.

Pulmonary aspergillosis in nonimmunocompromised hosts, although rare, is being increasingly recognized. The diagnosis of pulmonary aspergillosis is difficult, since the recovery of Aspergillus from respiratory samples cannot differentiate colonization from invasion. We assessed the role of bronchoalveolar lavage (BAL) in detecting galactomannan (GM) for diagnosing pulmonary aspergillosis in 73 nonimmunocompromised patients with pulmonary infiltrates for whom the test was ordered. Six patients had pulmonary aspergillosis, two each with acute invasive pulmonary aspergillosis, chronic necrotizing pulmonary aspergillosis, and aspergilloma. All six patients had a BAL GM level of >/=1.18. The sensitivity, specificity, and negative predictive value (NPV) for a BAL GM level of >/=1.0 were 100%, 88.1%, and 100%, respectively. Notably, the positive predictive value (PPV) was only 42.9%, likely reflecting the low prevalence of pulmonary aspergillosis among nonimmunosuppressed patients. The combination of BAL microscopy and culture had a sensitivity and NPV similar to those of BAL GM detection but a higher specificity and PPV (92.5% and 54.6%, respectively). Moreover, a BAL GM test did not identify any cases that were not diagnosed by conventional methods like microscopy and culture. In conclusion, there was no conclusive benefit of determining BAL GM levels in the diagnosis of pulmonary aspergillosis among nonimmunocompromised hosts. Given the likelihood of false-positive results, a BAL GM test should not be ordered routinely in this population.

Bronchoalveolar lavage galactomannan in diagnosis of invasive pulmonary aspergillosis among solid-organ transplant recipients.

We review the experience at our institution with galactomannan (GM) testing of bronchoalveolar lavage (BAL) fluid in the diagnosis of invasive pulmonary aspergillosis (IPA) among solid-organ transplant recipients. Among 81 patients for whom BAL GM testing was ordered (heart, 24; kidney, 22; liver, 19; lung, 16), there were five cases of proven or probable IPA. All five patients had BAL GM of > or = 2.1 and survived following antifungal therapy. The sensitivity, specificity, and positive and negative predictive values for BAL GM testing at a cutoff of > or = 1.0 were 100%, 90.8%, 41.7%, and 100%, respectively. The sensitivity of BAL GM testing was better than that of conventional tests such as serum GM or BAL cytology and culture. Moreover, a positive BAL GM test diagnosed IPA several days to 4 weeks before other methods for three patients. Twelve patients had BAL GM of > or = 0.5 but no evidence of IPA. Among these, lung transplant recipients accounted for 41.7% (5/12) of the false-positive results, reflecting frequent colonization of airways in this population. Excluding lung transplants, the specificity and positive predictive value for other solid-organ transplants increased to 92.9% and 62.5%, respectively (cutoff, > or = 1.0). In conclusion, BAL GM testing facilitated more-rapid diagnoses of IPA and the institution of antifungal therapy among non-lung solid-organ transplant recipients and helped to rule out IPA.

Pharmacoeconomic analysis of caspofungin versus liposomal amphotericin B as empirical antifungal therapy for neutropenic fever.

PURPOSE: An analysis was conducted that evaluated and compared the cost differences between caspofungin and liposomal amphotericin B when the medications were used as empirical antifungal therapy for persistent fever during neutropenia. METHODS: Rates of drug use and impaired renal function (IRF) were based on data from published studies. IRF was defined as a doubling of the serum creatinine level or, if the creatinine level was elevated at enrollment, an increase of at least 1 mg/dL. The estimates of the costs for drug acquisition and treating IRF were derived using published data and applied to compare caspofungin with liposomal amphotericin B. Sensitivity analyses were performed by varying the IRF and relative acquisition costs to assess the effect of these factors on the cost differences. RESULTS: The acquisition costs per patient were 6942 dollars for liposomal amphotericin B and 3996 dollars for caspofungin. The estimated cost per patient from IRF was 3173 dollars for liposomal amphotericin B and 793 dollars for caspofungin. Combining drug acquisition and IRF costs, the overall treatment cost per patient for caspofungin was 5326 dollars less than for liposomal amphotericin B. In sensitivity analyses of drug costs, the price of liposomal amphotericin B would have to be 23.95 dollars per vial for the overall treatment costs to be equal. CONCLUSION: Comparison of cost estimates derived from published data revealed that a combined estimate of acquisition costs and costs related to the treatment of IRF was lower for caspofungin than for liposomal amphotericin B for empirically treating patients with neutropenic fever.

Is combination antifungal therapy for invasive aspergillosis a necessity in hematopoietic stem-cell transplant recipients?

PURPOSE OF REVIEW: The use of combination antifungal therapy in hematopoietic stem-cell transplantation patients is controversial and limited by a paucity of controlled data. The recent literature is reviewed and the relative arguments for and against combination antifungal therapy are outlined with summative recommendations to assist practitioners in decision-making. RECENT FINDINGS: There is an abundance of in-vitro and murine in-vivo combination antifungal literature, whereas clinical data are less abundant and controlled. Of the published case series there is a suggested benefit to combination therapy over monotherapy, although there are limitations to the available literature. Other issues in the combination debate that are addressed include the following: improved response rates and a survival advantage have been demonstrated in recent monotherapy studies; response rates in most published combination therapy studies do not suggest large gains over monotherapy; the lack of sustained survival advantage to combination therapy studies; and finally the consideration of host defenses in treatment responses. SUMMARY: Based on available data, combination therapy is not warranted at the initial diagnosis of invasive aspergillosis. Randomized, controlled trials with rigorous study design are needed.

New strategies of antifungal therapy in hematopoietic stem cell transplant recipients and patients with hematological malignancies.

Invasive fungal infections (IFIs) are associated with considerable morbidity and mortality among high-risk individuals. Outcomes for IFI historically have been suboptimal and associated with a high mortality rate, hence global prophylaxis strategies have been applied to at-risk populations. Among certain populations, fluconazole prophylaxis has reduced systemic and superficial infections caused by Candida species. Newer azoles are currently being evaluated as prophylaxis and have the potential to provide protection against mould pathogens that are more troublesome to treat once they occur. Global prophylaxis strategies have the shortcoming of subjecting patients to therapy that ultimately will not need it. Targeted prophylaxis has the advantage of treating only patients at highest risk using some parameter of greater host susceptibility. Prophylaxis strategies are most suitable in patients at the highest risk for IFI. For patient groups whose risk is somewhat lower or when suspicion of IFI occurs in patients receiving prophylaxis, empirical antifungal therapy is often employed following a predefined period of fever. Again this approach subjects many non-infected patients to unnecessary and toxic therapy. A more refined approach such as presumptive or pre-emptive therapy whereby treatment is only initiated upon positive identification of a surrogate marker of infection in combination with clinical and radiological signs will subject fewer patients to toxic and expensive treatments.

Dose modification protocol using intravenous busulfan (Busulfex) and cyclophosphamide followed by autologous or allogeneic peripheral blood stem cell transplantation in patients with hematologic malignancies.

We evaluated the safety and toxicity through a 5-cohort dose-modification model of once-daily administration of IV busulfan (Bu) in combination with high-dose cyclophosphamide (Cy) as preparative therapy for stem cell transplantation. Twenty-one adult patients with hematologic malignancies were evaluated. Eleven patients underwent autologous and 10 patients underwent HLA-matched sibling allogeneic transplantation. Patients were sequentially enrolled into 5 cohorts. Cohort 1 received intravenous (IV) Bu 1.6 mg/kg every 12 hours for 2 doses and then 0.8 mg/kg every 6 hours for 12 doses; cohort 2 received IV Bu 1.6 mg/kg every 12 hours for 4 doses and then 0.8 mg/kg every 6 hours for 8 doses; cohort 3 received IV Bu 3.2 mg/kg for 1 dose and then 1.6 mg/kg every 12 hours for 2 doses and 0.8 mg/kg every 6 hours for 8 doses; cohort 4 received IV Bu 3.2 mg/kg every 24 hours for 2 doses and then 0.8 mg/kg every 6 hours for 8 doses; and cohort 5 received IV Bu 3.2 mg/kg every 24 hours for 4 doses. In all groups, Bu was administered on day -7 through day -4 and was followed at least 6 hours after the last Bu dose by Cy 60 mg/kg daily for 2 doses on days -3 and -2. Blood samples were collected for pharmacokinetic analysis on the first and last day of IV Bu administration. All patients were alive and had engrafted at day 30. Five patients developed grade 3 or 4 toxicities. Four patients developed hepatic abnormalities, and 3 exhibited evidence of veno-occlusive disease. Two of 3 patients in cohort 5 with a Bu area under the curve >6000 micromol/min developed autopsy-confirmed veno-occlusive disease. Interpatient variability in AUCs was observed in patients within and between cohorts, but no statistically significant interpatient differences were observed in Bu half-life, volume of distribution, clearance, or dose-adjusted area under the curve. Further, minimal variability in Bu pharmacokinetics was observed between the 2 evaluations performed in each patient, thus reflecting the stability of Bu disposition within individual patients. On the basis of the dosing guidelines and schedule outlined in this study, our data suggest that administration of IV Bu 3.2 mg/kg IV every 24 hours for 4 doses in combination with Cy may result in excessive toxicity.

Busulfan, cyclophosphamide, and etoposide as conditioning for autologous stem cell transplantation in multiple myeloma.

Autologous stem cell transplantation (ASCT) has enabled the use of high-dose alkylating agents either as a single agent or in combination with other cytotoxic agents and/or total body irradiation (TBI) for the treatment of multiple myeloma. Despite improved complete remission rates, relapse and regimen-related toxicities remain challenging. In an effort to increase event-free survival and decrease the high incidence of regimen-related toxicity, we have studied the use of etoposide in combination with reduced-dose busulfan and cyclophosphamide as a conditioning regimen for ASCT in a group of 26 patients with advanced multiple myeloma. Median follow-up for the group was 30 months. There was no early treatment-related mortality. The main toxicity was mucositis. Otherwise, there was 1 case of reversible, clinically diagnosed hepatic veno-occlusive disease. Post-engraftment, 10 patients (38%) achieved CR, 15 (58%) patients achieved PR or SD, and 1 patient developed progressive disease (4%). Five patients in PR and 1 with progressive disease before transplant attained a CR post-transplant. The median times for event-free survival and overall survival after transplantation were 24 and 43 months, respectively. In conclusion, conditioning with busulfan, cyclophosphamide, and etoposide followed by ASCT is a safe regimen with comparable effectiveness to other previously used preparative regimens, thus providing another approach of non-TBI containing high-dose chemotherapy for patients with multiple myeloma.

Prophylaxis, empirical therapy, or pre-emptive therapy of fungal infections in immunocompromised patients: which is better for whom?

Immunocompromised patients are at risk of developing fungal infections. Over time, the incidence of fungal infections and the spectrum of causative organisms have changed. In addition, treatment strategies in this high-risk population have also changed. Traditional approaches (using polyene-based therapy and older azoles), including empirical treatment strategies, have evolved to include prophylaxis in populations at the greatest risk. These strategies, although effective against Candida species, have not really impacted infections caused by Aspergillus spp. With the recent approval of antifungal agents with demonstrated activity against Aspergillus and other mould infections, there is hope for better outcomes in the treatment of established infections. Several agents, with activity against Aspergillus, have been shown to be effective in the empirical setting. The role of these new antifungal agents in the prophylactic setting remains unknown at present, but the potential for reducing Aspergillus infections is promising and requires ongoing study. The other area of significant research in fungal infections has been the search for accurate, non-invasive, rapid diagnostic tests. Over the past year, several publications have indicated that early diagnosis is possible in immunocompromised patients. These new diagnostics have paved the way for a new strategy, called pre-emptive therapy, enabling infected patients to be identified at an earlier stage of infection. This strategy will permit targeted antifungal therapy in those at greatest risk, and will avoid unnecessary, potentially toxic therapy in those not infected. Validations of the various techniques show promise and are reviewed in this paper.