SARS-CoV-2 Vaccinated Breakthrough Infections With Fatal and Critical Outcomes in the Department of Antioquia, Colombia

Andres Cardona One-Health Genomic Laboratory (Colombia/Wisconsin One-Health Consortium), Universidad Nacional de Colombia Simon Villegas One-Health Genomic Laboratory (Colombia/Wisconsin One-Health Consortium), Universidad Nacional de Colombia Maria Stella Lopez One-Health Genomic Laboratory (Colombia/Wisconsin One-Health Consortium), Universidad Nacional de Colombia Maria Angelica Maya Hospital Universitario de San Vicente Fundación Celeny Ortiz-Restrepo Gobernación de Antioquia, Medellin Olga Hernandez-Ortiz University of Antioquia Karl Ciuoderis One-Health Genomic Laboratory (Colombia/Wisconsin One-Health Consortium), Universidad Nacional de Colombia Laura Perez One-Health Genomic Laboratory (Colombia/Wisconsin One-Health Consortium), Universidad Nacional de Colombia Idabely Betancour Gobernación de Antioquia, Medellin Carolina Muñoz Gobernación de Antioquia, Medellin Lina Bustamante Gobernación de Antioquia, Medellin Andres Mauricio Rangel University of Antioquia Jorge Osorio University of Wisconsin–Madison Juan Pablo Hernandez-Ortiz (  jphernandezo@wisc.edu ) One-Health Genomic Laboratory (Colombia/Wisconsin One-Health Consortium), Universidad Nacional de Colombia Mark Anderson Abbott Laboratories, United States Francisco Averhoff Abbott Laboratories, United States Gavin Cloherty Abbott Laboratories, United States


Introduction
By September 10 th , 2021, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) had caused more than 226 million known cases of the coronavirus disease (Covid-19) worldwide, with an unfortunate outcome of 4.6 million deaths 1 . In response, an unprecedented effort was made to develop, authorize, and deploy vaccines, with more than 5,8 billion doses of several vaccines administered to date. These immunization strategies are mostly directed at the viral spike protein (S), but the emergence of viral variants, particularly of the S gene, threatens their continued e cacy to prevent infection and/or severe disease. Variants of concern (VOC) and variants of interest (VOI) have provided motivation to increase molecular testing and to perform viral genomic surveillance in infected persons, thereby fostering an understanding of the transmissibility and virulence of variants and their ability to evade current vaccines.
In Colombia, VOI and VOC were identi ed at the beginning of 2021, resulting in an extended wave of infections from March to July of 2021 (around 25 epidemiological weeks (EWs)). Previous infection peaks in Colombia lasted between 4 to 6 EWs. The circulating VOC and VOI include alpha (B.1.1.7), gamma  Figure 1, the introduction of emergent variants, the extended wave of infections, and the immunization program are contemporary events during the rst and second quarters of 2021. For completeness, the progress of the vaccination program in Antioquia is shown in the gure using the dates when increments of one million doses were reached.
Here, we describe 96 patients with breakthrough infections in the Departamento of Antioquia. Among them, 30 had a fatal outcome, 13 required ICU hospitalization, and 53 had mild or asymptomatic disease. Epidemiological, clinical, and viral-genomic characterizations of these patients are provided. The patients were identi ed within the COVID-19 diagnostic network of Antioquia, and the samples were processed for PCR testing and viral RNA sequencing in the One-Health Genomic Laboratory (OHGL), at the Universidad Nacional de Colombia, and in the Public Health Laboratory of Antioquia (PHLA). Our observations provide support for current strategies to monitor multiple variables proactively, including the determination of risk groups and the possible immune evasion of some variants.

Specimen collection
Samples (nasal swabs and nasopharyngeal washes) were collected through the COVID-19 diagnostic network of the Departamento de Antioquia, which is regulated and coordinated by the National Institute of Health (INS) and consists of 28 laboratories. The One-Health Genomic Laboratory (OHGL) and the Public Health Laboratory of Antioquia (PHLA) are part of this network, and it is also authorized and endorsed to perform genomic surveillance of SARS-CoV-2 by the INS. The standard operating procedures (SOPs) for sample collection and storage include careful custody of the samples at each laboratory with a mandatory storage at -80°C. Every week, criteria for genomic surveillance are determined and the corresponding samples are sent to the OHGL, where samples are stored at -80°C until processed. Typical criteria for genomic surveillance include hospitalized patients (general hospitalization and intensive care), a history of vaccination, travelers, randomly collected samples, and patients with suspected reinfection. During 2021, the OHGL had processed more than 1,700 samples, including approximately 200 from vaccinated patients. The COVID-19 immunization program in Antioquia currently uses CoronaVac (Sinovac), BNT162b2 (P zer-BioNTech), Ad26.COV2.S (Jansen) and ChAdOx1-S (AstraZeneca) vaccines. Epidemiological and clinical data were collected for each sample, i.e. age, gender, origin, date of onset of symptoms, date of sample collection, associated symptoms, comorbidities, and risk group.

PCR and Ct Calculation
RNA extraction was performed on swab samples and nasopharyngeal washings using the ZR viral kit (Zymo® research, Irvine, CA, USA) following manufacturer´s instructions. rRT-PCR was performed in a thermocycler CFX96 (Bio-Rad Laboratories, Inc, USA) using iTaq Universal Probes One-Step Kit (Biorad®) and speci c primers/probe targeting Envelope and RNA-dependent RNA polymerase (RdRP) genes 3 (Table S1). Running conditions were: one cycle at 50°C for 10 min, one cycle at 95°C for 3 min and 40 cycles of 95°C for 15 s and 58°C for 30 s. In addition, for each sample, an ampli cation of the Ribonuclease P gene (RNaseaP) was made as a control for the extraction of viral RNA. A Ct value was obtained from each of the samples according to the rRT-PCR curves; this value represents the cycle in which the samples have a positive ampli cation and exceed an RFU of 300 for the E and RdRP genes and of 100 for the Ribonuclease gene P. The samples were considered positive when the Ct value was less than 38.

Whole genome sequencing
Samples having a Ct below 27 were selected for sequencing using the nCoV-2019 sequencing protocol v3 4 . cDNA was rst made from the RNA (8 µl) and the LunaScript RT Supermix (5X) was used. Then, the multiplex PCR was performed using 218 primers, separated into 2 pools (pool 1 with 110 primers and pool 2 with 108 primers), and the Q5 hot start high delity 2X master mix was used (New England Biolabs, CITY, STATE). Two ampli cations were conducted for each sample (3 µl of cDNA) and subsequently pooled by mixing 5 ul of each PCR product and 40 ul of nuclease-free water. Each amplicon pool was quanti ed with the aid of the Qubit ™ dsDNA HS Assay Kit (Thermo Fisher Scienti c). The amplicon pools that had a concentration greater than or equal to 7 ng / ul were used for the end repair. This was conducted using between 50 and 100 ng of DNA and the NEBNext® Ultra ™ II End Repair / dA-Tailing Module (New England Biolabs). Then the barcode ligation was made using the Native Barcoding Expansion 96 kit (Oxford Nanopore) and the Blunt / TA Ligase Master Mix (New England Biolabs). Each sample was pooled and puri ed using AMPure XP beads (Beckman coulter). Next, the AMII was ligated using the Quick Ligation ™ Kit (New England Biolabs). DNA puri cation was performed again with the AMPure XP beads (Beckman coulter). Finally, the puri ed DNA was mixed with SQB buffer and Loading Beads (LB) and loaded into the Flow Cell (R9).
The assembly of raw NGS data was performed by following the pipeline described for Oxford Nanopore Technologies (ONT) platform 5 . Each sequence obtained was analyzed with the help of Pango-Lineage, which assigns a lineage according to the lineages reported so far 6 . In addition, Nextclade was used for the analysis of each of the mutations of the sequences 7 .

Results
Samples, along with clinical and epidemiological information, from patients and deceased persons with con rmed immunization (partial and complete) that are diagnosed as positive to SARS-CoV-2 by any laboratory of the Antioquia's network are constantly collected by the OHGL and the PHLA. A con rmatory RT-PCR test is carried out on the samples to qualify the Ct and validate the diagnosis. Special attention is made to deceased patients and individuals that required ICU hospitalization. Up to September 2021, we had detected 49,630 breakthrough infections in Antioquia, where 2998 required hospitalization, 1018 need an ICU and 502 died. From those patients more than 200 breakthrough infection samples had been recovered, and 96 had the quality to perform whole genome sequencing. From the latter, 30 patients had a fatal outcome, 13 required ICU, and 53 presented mild or asymptomatic disease. Patients were organized by age with No 1 being the oldest and No 96 the youngest ( Table 1). The Ct value of the RT-PCR was used to illustrate the impact of age and comorbidities on the outcome of the breakthrough infection (Figure 2A). The probability of getting an infection after n days post vaccination and comparisons for two groups: (i) deceased vs. surviving patients and (ii) Sinovac vs. P zer recipients, was also protted ( Figure 2B). Breakthrough infections of other vaccines were of insu cient quantity for analysis, not implying that they offer better protection, but a consequence of the low number of recipients (up to September 2021, Sinovac and P zer vaccines were administered to almost 80% of vaccinated recipients).
Analysis of deceased and surviving vaccine recipients with critical disease revealed that 72.09% of the patients had comorbidities and 44.17% were females. Advanced age and comorbidities were determinant factors in the outcome and progression of the disease (p < 0.01). The odds ratio (OR) indicates that it is 17.4 more probable to have a fatal outcome for patients older than 60 years of age, while it is 7.3 more probable to succumb from COVID-19 when comorbidities were present. The average age for deceased patients with comorbidities was 81, and 71 for deceased patients without known comorbidities. The average age of ICU survivors with comorbidities was 67, and 59 for ICU survivors without comorbidities. Patients with mild or asymptomatic infection averaged 49 years. The most common comorbidities in patients with fatal outcome were pulmonary arterial hypertension (PAH) (66.7%), diabetes (37.5%) and thyroid diseases (25%). On average these patients each had 3 comorbidities (Table 1). In contrast, survivors of critical disease presented with PAH (57.1%) and diabetes (57.1%) as the most prevalent comorbidities and had, on average, 1.5 comorbidities each ( Table 2). Accounting for the fact that the vaccination program in Colombia for adults older than 70 years used Sinovac in a high percentage, it is not surprising that this vaccine was more frequent in patients with fatal and ICU outcomes. Tables S2, S3 and S4 summarize information for deceased and ICU patients without comorbidities, patients with mild or asymptomatic disease and patients suspected of being reinfected (N = 15). Reinfected patients in this report were not con rmed by sequencing due to the poor quality of one of their samples. None of these patients had a fatal outcome and only one required an ICU hospitalization.

Discussion
The introduction of VOC and VOI in Colombia changed the dynamics of COVID-19 infection and generated extra burdens in non-clinical measures to control the ongoing pandemic. A major consequence of lineages with an increased reproduction number was an extended wave of infections that lasted for more than 4.5 months. The emergence of these variants might have not been exclusively responsible for this extended wave, as the country also experienced several episodes of civil unrest during this timeframe. As a consequence, standard social distancing measures, quarantines and other non-clinical measures were di cult to implement, however, it motivated the intensity of the vaccination programs, which started mid-February, and was initially concentrated in highrisk groups (advanced age, healthcare workers, and people with comorbidities). With 34.43% of the population fully vaccinated in the Departamento de Antioquia, the lethality of SARS-CoV-2 decreased from 2.2% in non-vaccinated patients to 0.014% in vaccinated patients 14 . The surveillance of patients with a breakthrough infection offers a mechanism to understand the protective capacity of existing vaccines against emerging variants. A total of 96 patient samples were characterized to provide crucial information and to delineate possible strategies to protect patients at risk. We found that advanced age and comorbidities were highly correlated with fatal and ICU outcomes in patients with breakthrough infections (p < 0.01). Concomitant with the natural course of the disease, PAH and diabetes were dominant in these patients 15,16 . Other comorbidities, including thyroid disease, cancer, coronary disease, chronic obstructive pulmonary disease (COPD) and chronic kidney disease (CKD) were present in approximately 25% of the patients. Correlated with the advanced age of these patients, around 40% presented neurological or mental disorders. The average age monotonically increased from patients with mild disease, to critical disease with and without comorbidities to deceased patients with and without comorbidities.
The probability of a breakthrough infection after n days post-vaccination was calculated from the distribution of infections from all patients. We found no correlation between the probability of infection and the outcome of disease, indicated by the similarities between the distributions and the cumulative probabilities in patients with fatal outcome and survivors. For instance, the probability to get infected after vaccination within the rst 30 days was 60% and 55% for deceased and survivors, respectively. On the other hand, signi cant differences are observed when the P zer-BioNTech and the Sinovac vaccines were compared. We found that the probability of a breakthrough infection with the P zer-BioNTech is higher within the rst 60 days after vaccination, i.e. 90% of the patients were infected between day 0 and day 60. The Sinovac vaccine, however, resulted in a atter distribution that persists for 120 days after vaccination.
These observations may indicate that the timeline for protection is, on average, faster for the P zer-BioNTech biological than for the Sinovac vaccine.
Examination of the SARS-CoV-2 sequences revealed that the mu variant was present in most of the patients with breakthrough infections with fatal and ICU outcomes. The gamma and the B.1.625 variants were also detected in a considerable number of patients. The elevated participation of the mu variant in these patients is not surprising according to recent reports regarding its diminished neutralization capacity in isolates 17 and pseudo-virus 18 . Accounting that we did not observe a dominant emerging variant, some of the breakthrough patients were also infected by persistent non-VOC/VOI variants (mostly of the B.1 group). Some substitutions of therapeutic and immunologic concern were prevalent in the majority of the variants, i.e. the E484K was observed in 90.1%, while the N501Y was present in 79.5%. Other substitutions, like the K417T and the L452R were present in 26.7% and 2.2% of the variants, respectively. Phylogenetic analysis determined the monophyletic nature of the VOC and VOI that circulate in Antioquia and deceased patients and ICU survivors are indicated in the corresponding clade of the variant (Figure 4). It also reinforces the neighboring similarities between the mu and the B.1.625 linage. In fact, in all sequenced mu and B.1.625 strains the substitutions T95I, E484K, D614G and D950N were present in the spike protein. Furthermore, a close relationship between mu and the other VOCs, such as delta, alpha, gamma, and lambda, was not observed. However, all share pivotal mutations in the Spike protein, such as E484K, D614G, P681H and D950N. These substitutions are of great interest because they have been related to higher rates of infection and the potential escape on the immune response 19,20,21,22 .
Fifteen of the patients with breakthrough infections were suspected of reinfection (Table S4). This classi cation was made merely on diagnosis, i.e., patients with two positive RT-PCR results with 90 days or more between. Up to September 2021, the OHGL and the PHLA have been unable to corroborate reinfection with sequencing. In most of the cases, one of the samples (either from the rst or second infection) did not meet the technical requirements for an acceptable sequence coverage. However, RT-PCR protocols and reagents in Colombia are standardized and the chances are high that the reinfected classi cation is appropriate. Among these 15 patients, no fatal outcomes were observed and only one patient required an ICU hospitalization. The other patients had mild or asymptomatic disease. The identi ed viral linages were gamma (7), mu (4), B.1.625 (2) and parental (2).
Our observations empower the importance of the urgent efforts being taken globally to achieve herd immunity and vaccinate most of the global population. They also validate the effectiveness of the vaccines and highlights the need to stay vigilant and monitor viral evolution and its potential impact on vaccine e cacy. As many countries are preparing themselves for re-immunization campaigns, vaccine manufacturers might also consider updating their formulation to provide coverage against emerging variants. In addition, they leverage the necessity of universal vaccines against beta-coronaviruses and highlight the importance of active surveillance as a platform for pathogen discovery. More importantly, our observations show the importance of the ongoing race between immunization and the potential emergence of viral escape mutants. Our analysis similarly supports the need to maintain molecular testing (in symptomatic, asymptomatic, and vaccinated patients) and viral genome surveillance to establish databases for breakthrough infections.

Declarations
Ethics Statement. The study was approved by the Institutional Review Board of the Corporación de Investigaciones Biológicas (CIB) in Medellin, Colombia, and by the Secretaria Seccional de Salud de Antioquia. Informed consent was obtained from all subjects or family members. Anonymized clinical and epidemiological data was collected and used according to the guidelines presented to the ethical committee. Genomic information was used after clearance Patient à  2  3  4  5  6  8  9  12  13  14  15  16  18  19  20  21  26  27  28  29  33  40  55   Age  à  96  95  93  92  91  91  88  87  87  86  85  85  84  84  83  83  81  81  80  80  77  73  63   Gender à  F  F  M  M  F  M  F  M  F  F  F  M  M  F  M  M  M  F  M  F  M  M  F   Vaccine à  sv  sv  sv  sv  sv  sv  sv  sv  sv  sv  sv  sv  sv  pz  sv  sv  sv  sv  sv  sv  sv  pz  sv   Partial Vaccination à

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. BIVaccinatedSI.docx