RADIOIMMUNOTHERAPY (RIT) VERSUS AUTOLOGOUS HEMATOPOIETIC STEM-CELL TRANSPLANTATION (ASCT) IN RELAPSED/REFRACTORY (R/R) FOLLICULAR LYMPHOMA: A FONDAZIONE ITALIANA LINFOMI (FIL) PHASE III TRIAL

Background: Optimal consolidation for young R/R FL in the rituximab age remains uncertain and the benefit of ASCT is not clearly established. Aims: The FIL FLAZ12 trial (NCT01827605) is a prospective, multicenter, randomized, phase 3 trial, comparing RIT versus ASCT, as consolidation after chemoimmunotherapy, both followed by R maintenance in R/R FL. Methods: Pts aged 18-65 yrs, with R/R FL after 1 or 2 lines of chemoimmunotherapy, without significant comorbidities were enrolled. Patients received 3 courses of therapy chosen by the investigator among RCHOP, R-DHAP, R-FM, R-ICE, R-IEV or R-B. Pts achieving at least PR (according to Cheson et al. 2007) were randomized 1:1 to either RIT or ASCT before CD34+ collection. Conditioning for ASCT was BEAM or TEAM. RIT was given as previously described (Morschhauser et al., 2008). After consolidation, pts received R maintenance every 3 months for eight courses. Primary endpoint was PFS. Considering ASCT toxicity, it was hypothesized to be a superior choice, if capable of increasing 3-years PFS from 40% to 60% (two-side log-rank test with alpha of 5% and a power of 85%). Clinical secondary endpoints were ORR, CRR, OS, EFS and TTF. Results: Between Aug 2012, and Sep 2019, 164 pts were screened and 159 enrolled by 38 FIL Centers (enrolled population). Unfortunately, the study was prematurely closed due to low accrual. The data were analyzed on an ITT basis on May 2, 2021 with a median follow-up (mFU) from enrollment of 43 months and 75 PFS events. The two arms were clinically well balanced, with median age of 57 yrs (IQR 49-62), 55% male, 57% stage IV, 20% bulky disease. Tumor re-biopsy was performed in 79% pts. POD-24, retrospectively assessed was observed in 32% of pts. Two pts (1%) did not start treatment (non-confirmed histology and withdrawal). Sixteen (10%) pts discontinued before randomization (7 SD, 3 PD, 3 AE, 1 withdrawal, 2 poor compliance) and 141 (89%) were randomized to either RIT (71) or ASCT (70) (randomized population). Of these 19 (13%) (RIT 8, ASCT 11) did not receive the planned consolidation due to 7 PD, 4 AE, 1 medical decision, 2 poor mobilization, 2 withdrawals, 1 poor compliance, 2 protocol breaches, while 63 (89%) received RIT and 59 (84%) ASCT. After RIT, 61% of pts achieved CR and 23% PR, while after ASCT these were 70% and 9%. Estimated PFS at 3 yrs was 60% (95% CI: 46%-71%) in the RIT arm vs. 59% (95% CI:45%-70%) in the ASCT arm, p = 0.8613 (HR 0.96, 95%CI: 0.57,1.59). (Figure 1) 3yrs-OS was again superimposable in the two arms: 83% (95%CI: 69%-91%) in the RIT vs 85% (95% CI: 72%-91%) in the ASCT, p = 0.8310 (HR 1.10, 95%CI: 0.45,2.72). Grade ≥ 3 hematological toxicity was 46% in the RIT vs 94% in the ASCT arm (p < 0.001). For ASCT vs RIT grade ≥3 neutropenia occurred in 94% vs 41% of pts (p < 0.001). During follow-up, 4 pts died in remission: 1 AML (RIT), 2 SARS-COV2 infections (RIT) and 1 pneumonia (ASCT). Second cancers occurred in 3 pts after RIT and 7 after ASCT (p = 0.480). Multivariable analysis for PFS indicated POD-24, male sex, LDH and refractory disease as adverse parameters. Subgroup analysis for PFS including gender, age, LDH, POD-24 and extranodal disease show no subgroup favoring RIT nor ASCT. Image: Summary/Conclusion: Even if prematurely interrupted, our study demonstrated no meaningful difference in efficacy between ASCT and RIT, but ASCT was more toxic and more demanding for pts and health service. Both strategies induced a similar and favorable long-term outcome suggesting that consolidation programs milder than ASCT require further investigation in R/R FL.

Long-term benefit of lurbinectedin as palliative chemotherapy in progressive malignant pleural mesothelioma (MPM): final efficacy and translational data of the SAKK 17/16 study.

BACKGROUND: The SAKK 17/16 study showed promising efficacy data with lurbinectedin as second- or third-line palliative therapy in malignant pleural mesothelioma. Here, we evaluated long-term outcome and analyzed the impact of lurbinectedin monotherapy on the tumor microenvironment at the cellular and molecular level to predict outcomes. MATERIAL AND METHODS: Forty-two patients were treated with lurbinectedin in this single-arm study. Twenty-nine samples were available at baseline, and seven additional matched samples at day one of cycle two of treatment. Survival curves and rates between groups were compared using the log-rank test and Kaplan-Meier method. Statistical significance was set at P value <0.05. RESULTS: Updated median overall survival (OS) was slightly increased to 11.5 months [95% confidence interval (CI) 8.8-13.8 months]. Thirty-six patients (85%) had died. The OS rate at 12 and 18 months was 47% (95% CI 32.1% to 61.6%) and 31% (95% CI 17.8% to 45.0%), respectively. Median progression-free survival was 4.1 months (95% CI 2.6-5.5 months). No new safety signals were observed. Patients with lower frequencies of regulatory T cells, as well as lower tumor-associated macrophages (TAMs) at baseline, had a better OS. Comparing matched biopsies, a decrease of M2 macrophages was observed in five out of seven patients after exposure to lurbinectedin, and two out of four patients showed increased CD8+ T-cell infiltrates in tumor. DISCUSSION: Lurbinectedin continues to be active in patients with progressing malignant pleural mesothelioma. According to our very small sample size, we hypothesize that baseline TAMs and regulatory T cells are associated with survival. Lurbinectedin seems to inhibit conversion of TAMs to M2 phenotype in humans.

P.0279 Neural underpinnings of depressive and post-traumatic symptomatology in covid-19 survivors: a voxel-based morphometry study

Background: A high prevalence of depression, anxiety, insomnia and PTSD has been reported in COVID-19 survivors [1]. This is similar to what previously observed in other Coronavirus-related diseases such as SARS and MERS [2]. The pathophysiology of post-infection neuropsychiatric symptoms is likely to be multifactorial, with a role played by inflammatory and immunological factors [3], but it is still largely unknown;we thus investigated COVID-19 survivors via 3T MRI imaging to identify neural underpinnings of post-infection neuropsychiatric symptoms in order to further elucidate their complex pathophysiology. Methods: Covid-19 survivors were recruited during an ongoing prospective cohort study at IRCCS San Raffaele Hospital in Milan;psychopathology was initially measured via several self-report questionnaires (Impact of Events Scale-Revised (IES-R), Zung Self-Rating Depression Scale (ZSDS), 13-item Beck's Depression Inventory (BDI));subsequently patients (n=28) underwent 3T MRI scanning (Philips 3T Ingenia CX scanner with 32-channel sensitivity encoding SENSE head coil). T1 weighted images were processed using Computational Anatomy Toolbox (CAT12) for Statistical Parametric Mapping 12 (SPM12) in Matlab R2016b;segmentation into Gray Matter, White Matter and cerebrospinal fluid, bias regularization, non-linear modulation and normalization to MNI space were performed;measures of Total Intracranial Volume (TIV) were obtained and images were smoothed with an 8-mm full width at half maximum Gaussian filter. Multiple regressions were performed using SPM12 software package: with no a priori regions of interest selected, whole-brain gray matter volumes were used as dependent variables, psychometric scales scores as independent variables, and age, sex and TIV as nuisance covariates. Results: After VBM regression analysis covarying for age, sex and TIV, ZSDS Index scores were inversely correlated with gray matter volume in the Bilateral Anterior Cingulate Cortex (MNI 2, 24, 28, cluster level pFWE = 0.045, k=767);furthermore 3 cluster were identified comprising again the anterior cingulate cortex and the insular cortex bilaterally in which IES-R scores were inversely correlated with gray matter volumes (Cluster 1: MNI -30, 9, 3, cluster level pFWE = 0.005, k=1284;Cluster 2: MNI 36, -3, -3, cluster level pFWE = 0.037, k=773;Cluster 3: MNI 9, 30, 28, cluster level pFWE = 0.038, k=766). No other statistical significant result was found. Conclusions: Our study identified an inverse correlation between anterior cingulate cortex volumes and depressive symptomatology, measured via ZSDS, and between bilateral insulae and anterior cingulate cortex volumes and the degree of distress in response to the traumatic event, measured via the IES-R. Analogous findings have already been reported in patients with Major depression [4] and PTSD [5], and our study confirms the role of volumetric reductions of these brain regions in depressive and post-traumatic symptomatology. Given the nature of our study it is not possible to infer whether the reduction of gray matter volume is a consequence of the Covid-19 infection itself or, as it appears more likely, precede the infection acting as predisposing factor for the subsequent development of depressive and post-traumatic symptomatology. No conflict of interest

P.0691 Mood-congruent cognitive distortion and processing bias in depressed covid-19 survivors: a comparison with major depressive disorder

Introduction. COVID-19 survivors often experience psychiatric sequelae, with depressive psychopathology as the leading cause for needing psychiatric intervention [1]. Depressive cognitive distortion is a core feature of major depression, fostering the experience of negative emotions and hampering recovery [2]. Moreover, cognitive biases are well-documented in patients with inflammatory diseases and associated depressive symptomatology [3]. Considering both the high prevalence of clinical depression among COVID-19 survivors and the critical role of cognitive distortions in depression, we consider of crucial importance to investigate cognitive processing biases in COVID-19 survivors. Methods. We studied 729 participants, divided in three groups: (1) 362 COVID-19 survivors;(2) 73 inpatients with Major Depressive Disorder (MDD);(3) 294 healthy participants (HC). Severity of depression was self-rated on the Zung Self-Rating Depression Scale (ZSDS). Neuropsychological bias toward emotional stimuli and the general negative outlook on the self were tested in a self-description task [4], during which subjects were asked to self-attribute or refuse positive and negative morally tuned adjectives, and latencies and frequencies of attribution were recorded. Depressive dysfunctional attitudes in causal attribution and interpretation of hypothetical events were measured on the Cognition Questionnaire (CQ). We performed homogeneity of slope or separate slopes analysis when appropriate in the context of Generalized linear model (GLMZ), with an identity link function. Likelihood ratio test was computed as a measure of significance for tested effects and Akaike Information Criterion (AIC) was obtained as goodness of fit measure [5]. Results. 22.4% COVID survivors self-rated their depressive symptoms above the clinical threshold. Bias in speed of information processing significantly predicts self-description in all groups (COVID depressed: Wald W2=19.81;COVID non depressed: W2=15.48;MD: W2=13.65;HC: W2=33.54;all p<0.001). Information processing bias and frequencies of attribution of morally negative elements strongly predicted the severity of self-rated depressive psychopathology (ZSDS scores) (Processing bias: LR χ2=40.99, p<0.0001;Frequencies: LR χ2=127.89, p<0.0001). Additionally, the cognitive distortion in causal attribution and interpretation of hypothetical events (CQ scores) in depressed post-COVID patients showed intermediate levels of severity in all dimensions between non-depressed post-COVID patients, and MDD (post-hoc Fisher's least significance test: p<0.05 at all comparisons). Moreover, the CQ total score significantly influenced the ZSDS scores (χ2=84.60, p<0.0001). Interestingly, homogeneity of slope analysis revealed regression slopes were parallel in COVID-depressed and hospitalized MD patient groups in all models, yielding no significant group interaction. Finally, bias in information processing and negative self-description both predicted CQ scores (Latencies ratio: χ2=3.91, p=0.0479;Frequencies: χ2=42.96, p<0.0001). Conclusions. The breadth of moral self-reproach and the severity of cognitive distortion in evaluating events showed the same association with severity of depression in MDD and in post-COVID depressed patients, distributing along a gradient of severity, thus suggesting that these individual features of depressive cognitive distortion are shared in these conditions and should be addressed as treatment targets in depressed COVID-19 survivors. No conflict of interest

P.0404 Rapid antidepressant response to first-line selective serotonin reuptake inhibitors in post-COVID-19 depression

Introduction: Depression was reported in 30–40% of patients at one, three, and six months following COVID-19 [1]. The host immune response to SARS-CoV-2 infection and related severe systemic inflammation seems to be the main mechanism contributing to the development of post-COVID depression. Emerging literature suggests anti-inflammatory and antiviral properties of antidepressants in the treatment of SARS-CoV-2 infection [2]. We hypothesized that post-COVID depression, triggered by infection and sustained by systemic inflammation, could particularly benefit from antidepressants. Thus, the present study aims to investigate the efficacy of SSRI in treating post-COVID depression. Methods: We included 58 adults patients who showed depressive episodes in the six months following COVID-19. We excluded patients if they showed: other psychiatric comorbidities, ongoing treatment with antidepressants or neuroleptics, somatic disease and medications known to affect mood. The severity of depression was rated at baseline and after for four weeks from the start of the treatment on the Hamilton Depression Rating Scale (HDRS) and response was considered when the patients achieved a 50% HDRS reduction after treatment. Statistical analyses to compare group means and frequencies (Student's t-test, Pearson χ2 test) were performed. To investigate changes in HDRS scores over time, repeated measures ANOVAs (according to sex, mood disorder history, and antidepressant molecule) were performed. Results: We found that 53 (91%) patients showed a clinical response to antidepressant treatment. Age, sex, mood disorder history, and hospitalization for COVID did not affect the response rate. Patients were treated with sertraline (n=26), citalopram (n=18), paroxetine (n=8), fluvoxamine (n=4), and fluoxetine (n=2). From baseline to follow-up, patients showed a significant decrease over time of HDRS score (F=618.90, p<0.001), irrespectively of sex (0.28, p=0.599), mood disorder history (F=0.04, p=0.834), and drug used (F=1.47, p=0.239). Discussion: Common knowledge highlights that among antidepressant-treated patients, response rates are moderate (40–60%). On the contrary, we observed a rapid response to the first-line antidepressants in more than 90% of patients irrespectively of clinical variables, thus suggesting a higher antidepressant response rate in post-COVID depression. The pathophysiology of post-COVID neuropsychiatric sequelae mainly entails severe systemic inflammation and subsequent neuroinflammation. In this context, we have previously found that one and three months after COVID-19, the severity of depression was predicted by the baseline systemic immune-inflammation index (SII) [3,4]. Furthermore, we found a protective effect of the IL-1β and IL-6 receptor antagonist against post-COVID depression possibly associated with their effect in dampening SII [5]. Mounting evidence suggests that antidepressants may a) decrease markers of inflammation;b) may inhibit acid sphingomyelinase preventing the infection of epithelial cells with SARS-CoV-2;c) may prevent the COVID-19 related cytokine storm by stimulating the σ-1 receptor;d) may exert antiviral effects via lysosomotropic properties;e) may inhibit platelets activation [2]. In conclusion, we hypothesized that post-COVID depression could particularly benefit from antidepressants since this molecules have anti-inflammatory and antiviral properties, pass the BBB and accumulate in the CNS, thus preventing the neuro-inflammation triggered by SARS-CoV-2 and associated with post-COVID depression. No conflict of interest

P.0086 Clinical and psychopathological predictors of fatigue in COVID-19 survivors: a machine learning study

Introduction: The COVID-19 pandemic has led to profound mental health consequences observed during acute infection and at short, medium, and long-term follow-up [1–3]. When considering long-term sequelae, a prevalent proportion of patients infected by SARS-CoV-2 experience a “Post-COVID-19 Syndrome” characterized by fatigue, depressive symptoms, sleep disturbances, and myalgia. In this context, fatigue is recognized as one of the leading complaints in COVID-19 survivors [4]. Long-term health consequences following COVID-19 and their impact on daily quality of life are largely unknown and need further investigation. Thus, questions about possible effects of mental health on fatigue, and of COVID-19 clinical severity on both, remained unanswered. We aim to predict long-term fatigue symptoms basing on clinical and psychopathological predictors through a machine learning approach. Methods: We evaluated the fatigue syndrome and the psychopathological status of 122 adult COVID-19 survivors (80 male, mean age 59.8±12.9) six months after hospital discharge for COVID-19. Clinical and psychopathological predictors were collected for the entire sample. Fatigue at six months was assessed using the Fatigue Severity Scale (FSS). Descriptive statistical analyses to compare means and frequencies were performed. To better disentangle the relationship between somatic and psychopathological predictors and the development of fatigue, we explored the effect of each predictor in affecting fatigue by implementing 5000 non-parametric bootstraps enhanced elastic net penalized logistic regression. The model's accuracy was estimated by 5-folds stratified nested cross-validations in the outer loop to define balance accuracy value (BA), class accuracies, and area under the receiver operator curve (AUC) (for a complete description of the method see [5]). Results: Six months after hospital discharge, 28%, 29%, and 24% of the total sample showed respectively depression (according to Zung Self-Rating Depression Scale), anxiety (according to State-Trait Anxiety Inventory form Y), and sleep disturbances (according to Women's Health Initiative Insomnia Rating Scale). Fatigue was present in 19% of the patients. When entering demographical, clinical, and psychopathological predictors in the elastic net penalized logistic regression, only depressive symptomatology significantly predicted the presence of fatigue at six months (Log Odds Ratio: 2.33;Standard deviation: 1.58;Lower and Upper 95% CI: -0.78 - 5.43;Variable Inclusion Probability: 96.7%). The 10-folds cross-validated elastic net model predicted fatigue with a BA of 65%, an AUC of 77%, and a specificity for the absence of fatigue of 74%, and a sensitivity for the presence of fatigue of 55%, showing good performances in excluding fatigue syndrome. Discussion: Besides confirming a high rate of long-term neuropsychiatric sequelae, our main finding is the strict association between fatigue and depression. We fear that, rather independent of pneumonia severity, major depression after COVID-19 is associated with persistent fatigue, thus worsening the burden of a non-communicable condition triggered by infection and by infection-related systemic inflammation, but then persisting on its own. Post-COVID syndrome, mainly characterized by fatigue, depression, and sleep disturbances, will affect COVID-19 survivors' daily functioning and place additional burden on the healthcare system. Clarifying the mechanisms and risk factors underlying such long-term symptomatology is essential to identify target population and to tailor specific treatment and rehabilitation interventions to foster recovery. No conflict of interest

P.0267 Persistent psychopathology in covid-19 survivors at one-year follow-up

Introduction: The effects of COVID-19 are highly variable, with potential involvement of almost all organs and systems. While the acute and sub-acute symptoms have been well described, the possible long-term sequelae of COVID-19 have become an increasing concern [1]. One, three, and six-months follow-up studies have reported highly prevalent post COVID neuropsychiatric sequelae [2,3,4,5]. The aim of the present study is to investigate the psychopathological impact of COVID-19 in survivors at one-year follow-up, also considering the effect of possible risk factors. Methods: We prospectively evaluated the psychopathological status of 160 COVID-19 survivors one year after hospital discharge during an ongoing prospective cohort study. To keep a naturalistic study design, exclusion criteria were limited to patients under 18 years. Sociodemographic and clinical data were collected. Current psychopathology was measured using the following self-report questionnaire: Zung Self-Rating Depression Scale (ZSDS), Impact of Events Scale-Revised (IES-R), State-Trait Anxiety Inventory form Y (STAI-Y), and Fatigue Severity Score (FSS). Need of antidepressant or anxiolytic treatment in the last year was collected. Statistical analyses to compare group means and frequencies (Student's t-test, Pearson χ2 test) exploring effects of sex, psychiatric history, and hospitalization for COVID-19 were performed. Results: Overall, 77 patients (48%) scored in the clinical range in at least one psychopathological dimension among depression, anxiety, and PTSD. Females and patients with a positive previous psychiatric diagnosis showed an increased score on most measures (Table). Hospitalization for COVID-19 did not affect psychopathology. During the year after COVID-19, 25 (16%) and 23 (14%) patients started an antidepressant or anxiolytic treatment respectively.Discussion: This is the first study that investigates psychopathology in a sample of COVID-19 survivors at one-year follow-up after hospital treatment. We reported high rates of persistent psychopathology consistently with previous coronavirus outbreaks. Psychiatric consequences to SARS-CoV-2 infection can be caused by the immune-inflammatory response to the virus itself or by psychological stressors such as social isolation, concerns about infecting others, and stigma. Considering that neuropsychiatric sequelae associates with a markedly increased risk of all-cause mortality, and given the alarming prevalence of post-COVID psychopathology, we now suggest to routinely asses psychopathology of COVID-19 survivors in order to promptly diagnose emergent disorders and to treat them to reduce the disease burden and related years of life lived with disability. No conflict of interest

An international registry on COVID-19 related hyperinflammation in children and young adults (HyperPED-COVID)

Introduction: By now, most of the children with COVID-19 infection have only mild symptoms. However, in the past year a small number of children have developed a serious inflammatory condition in temporal association with COVID-19 pandemic. This condition, named Multisystem Inflammatory Syndrome in Children (MIS-C), is characterized by a systemic inflammation with multiorgan failure;the clinical and laboratory features are similar to those usually observed in Kawasaki disease, cytokine storm syndrome and macrophage activation syndrome. While the SARS-CoV-2 PCR on nasal swab is positive only in a minority of patients, the serology is positive in most of them. Objectives: to create an International multicenter collection of patients with MIS-C involving the main pediatric networks committed in the care of patients with hyperinflammatory conditions. Primary endpoint of the project is to collect information on clinical presentation, laboratory parameters, clinical outcome and response to treatment of patients with MIS-C. Secondary endpoints of the project are: 1) to analyse different clusters in clinical phenotype in relation to age and geographical location, 2) to identify clinical and laboratory predictors of disease severity and outcome, 3) to evaluate the availability of samples of patients in the repositories linked to the different centers. Methods: a steering committee constituted by representatives of ERN-RITA, PRES, ESID and ISSAID and with the coordination of PRINTO developed a shared form to collect clinical manifestations, laboratory features, response to treatment and outcome of patients with MIS-C. The registry is available online on PRINTO website (www.printo.it). Results: a first survey between centers members of PRINTO network identified 365 patients from 51 centers (43 countries). Currently, 180 patients have been enrolled in the registry from 25 centers worldwide. Moreover, 11 additional centers have been activated and are ready to start enrollement. Ethical committee submission or activation procedures are ongoing in 36 more centers. 31 European countries and 12 extra-European countries are actively involved. Conclusion: After some interesting national initiatives, the HyperPEDCOVID registry will give the chance to analyze the impact of Multisystem Inflammatory Syndrome Children at the European and extra- European level, giving the possibility to analyze the distribution, clinical presentation and long-term outcome of this condition in different countries. The registry can also represent the basis for further biological and genetic studies in these patients.

An inflammatory profile correlates with decreased frequency of cytotoxic cells in coronavirus disease 2019. (COVID special issue #2.)

Increased production of inflammatory cytokines and myeloid-derived suppressor cells occurs in patients with coronavirus disease 2019. These inversely correlated with perforin-expressing natural killer (NK) and CD3⁺ T cells. We observed a lower number of perforin-expressing NK cells in intensive care unit (ICU) patients compared with non-ICU patients, suggesting an impairment of the immune cytotoxic arm as a pathogenic mechanism.