OS based on product-limit (Kaplan-Meier) estimates. Confidence intervals for the median are calculated according to Brookmeyer and Crowley.

If a subject had not died, survival time was censored at the last date the subject was known to be alive.

The analysis population was the intent-to-treat (ITT) subpopulation, which includes all subjects who were randomized to investigational medicinal product (IMP). Analyses performed on the ITT analysis set will take into account subjects' allocation to treatment groups as randomized and not as treated.

Tumor assessments were done via computed tomography (CT) or magnetic resonance imaging (MRI) scans and evaluated per RECIST v1.1. The assessment for measurable disease during screening (within 14 days prior to Day 1) acts as the baseline assessment. Best OR was summarized for each treatment group by means of counts and percentages for the following categories: Complete Response (CR: disappearance of all target lesions), Partial Response (PR: at least a 30% decrease in the sum of diameters of target lesions), Progressive Disease (PD: at least a 20% increase in the sum of diameters of target lesions), Stable Disease (SD: neither sufficient shrinkage to qualify for PR nor sufficient increase to qualify for PD) or Not Evaluable (NE).

The analysis population was the ITT subpopulation, which includes all subjects who were randomized to IMP. Analyses performed on the ITT analysis set will take into account subjects' allocation to treatment groups as randomized and not as treated.

PFS based on product-limit (Kaplan-Meier) estimates. Confidence intervals for the median are calculated according to Brookmeyer and Crowley. Death will only be considered as an event if it occurs within 12 weeks after last tumor response assessment without progression.

The analysis population was the ITT subpopulation, which includes all subjects who were randomized to IMP. Analyses performed on the ITT analysis set will take into account subjects' allocation to treatment groups as randomized and not as treated.

TTF based on product-limit (Kaplan-Meier) estimates. Confidence intervals for the median are calculated according to Brookmeyer and Crowley.

The analysis population was the ITT subpopulation, which includes all subjects who were randomized to IMP. Analyses performed on the ITT analysis set will take into account subjects' allocation to treatment groups as randomized and not as treated. Subjects who were randomized but not treated have been censored at the date of randomization.

AEs were coded according to the Medical Dictionary for Regulatory Activities (MedDRA) classification. The incidence and type of AEs (i.e., serious AE [SAE], treatment-emergent AE [TEAE]) were summarized by dose cohort according to MedDRA system organ classes and preferred terms. An AE was considered as treatment-emergent if it occurred during or after the first IMP administration. An AE that occurred before the first IMP administration and worsened thereafter was also considered an AE. Worsening was reported as a new AE.

The analysis population was the safety analysis subpopulation, which includes all subjects who were administered any dose of IMP, and in addition those subjects in Arm C for which the intended control treatment is BSC. Subjects will be analyzed as treated and not as randomized.

Treatment duration (weeks) is calculated as [(last dose date of Sym004 - first dose date of Sym004)+7] / 7 days.

Sym004 dose received (mg/kg) is calculated as (total dose administered (mg)/weight (kg)).

Dose Intensity is calculated as (cumulative Sym004 dose (mg/kg) / treatment duration (weeks)).

Relative Dose Intensity is calculated as (dose intensity / planned dose intensity at randomization)*100.

Percentages are based on the number of subjects in the safety analysis set.

The analysis population was the safety analysis subpopulation, which includes all subjects who were administered any dose of IMP. Subjects will be analyzed as treated and not as randomized.

The Sym004 serum concentration used for the PK evaluation was calculated as the sum of the serum concentrations of the 2 component monoclonal antibodies of Sym004 (futuximab and modotuximab).

Trough Concentration (Ctrough) is equivalent to the concentration collected at the pre-dose timepoint.

Maximum Concentration (Cmax) is equivalent to the concentration collected at the end of infusion (EOI) timepoint.

The analysis population was the PK analysis set. Bioanalysis for serum concentration was done for a subset of subjects (N=19) at all scheduled timepoints; it was carried out only at Weeks 3, 5, 7 and the End of Treatment visit for all other subjects. Additionally, the Week 1 Day 1 EOI concentration for Subject 2740012 was assessed.

Tmax was defined as the time the PK sample was taken at end of infusion (EOI) relative to the start time of infusion (i.e., time between the start of infusion and the time of the EOI sample). For presentation of individual PK parameters and calculation of mean parameters, half of the lower limit of quantitation (LLOQ) value was used for concentration values below the LLOQ. The Sym004 serum concentration used for the PK evaluation was calculated as the sum of the serum concentrations of the 2 component monoclonal antibodies of Sym004, futuximab and modotuximab.

The analysis population was the PK analysis set, defined as subjects having at least 1 Sym004 serum concentration above the LLOQ. Exposure to Sym004 was confirmed in the majority of subjects treated with Sym004 for at least 1 timepoint post-dose.

A validated double antigen bridging ELISA was used for screening, confirmation, and titration of patient samples for anti-Sym004 ADA. Using rabbit anti-Sym004 as an ADA control antibody, the lower limit of detection was 54 ng/mL in the absence of Sym004 and 500 ng/mL in the presence of Sym004 at 5 µg/mL The timepoints for ADA sampling were chosen by the original sponsor for this trial. After the trial was transferred to Symphogen A/S, it was determined that not all samples were necessary for analysis. This is why the collection time points specified in the Outcome Measure Time Frame do not match with the Outcome Measure Data Table.

The analysis population was the safety analysis subpopulation, which includes all subjects who were administered any dose of IMP. Subjects will be analyzed as treated and not as randomized.

Scale: European Organisation for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire (Version 3) [QLQ-C30, Version 3].

The QLQ-C30 is a 30-question scale used to assess cancer patients' quality of life based on 15 factors (e.g., global health status, physical functioning, role functioning, etc.). The scale is composed of both multi-item scales and single-item measures. All of the scales and single-item measures range in score from 0 to 100:

• A high score for a functional scale represents a healthy level of functioning.
• A high score for the global health status represents a high quality of life.
• A high score for a symptom scale/item represents a high level of symptomatology (problems).

This measure was self-reported. Numbers analyzed between Week 1 and Week 7 differ from each other, as well from the overall number of subjects analyzed. Data could not be collected from subjects not compliant with reporting or once discontinued.

Scale: European Organisation for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire - Colorectal Cancer Module (QLQ-CR29).

The QLQ-CR29 is a 29-question scale used to assess colorectal cancer patients' quality of life based on 22 factors (e.g., body image, anxiety, weight, etc.). The scale is composed of both multi-item scales and single-item measures. All of the scales and single-item measures range in score from 0 to 100:

• A high score for a functional scale/item represents an unhealthy level of functioning, with the exception of one (1) scale pertaining to sexual interest (separated by sex).
• A high score for a symptom scale/item represents a high level of symptomatology (problems).

This measure was self-reported. Numbers analyzed between Week 1 and Week 7 differ from each other, as well from the overall number of subjects analyzed. Data could not be collected from subjects not compliant with reporting or once discontinued.

Scale: Functional Assessment of Cancer Therapy-Epidermal Growth Factor Receptor Inhibitor 18 (FACT-EGFRI-18).

The FACT-EGFRI-18 is an 18-question scale used to assess EGFR-inhibitor-treated cancer patients' quality of life relative to their experience of skin rash based on three (3) multi-item subscales. The subscales combined (i.e., Symptom Index) range in score from 0 to 72. A higher score represents a high level of symptomatology (problems).

High scores for all subscales represent a worse outcome:

• The Physical subscale ranges in score from 0 to 28.
• The Social/Emotional subscale ranges in score from 0 to 24.
• The Functional subscale ranges in score from 0 to 20.

This measure was self-reported. Numbers analyzed between Week 1 and Week 4 differ from each other, as well from the overall number of subjects analyzed. Data could not be collected from subjects not compliant with reporting or once discontinued.

OS based on product-limit (Kaplan-Meier) estimates. Confidence intervals for the median are calculated according to Brookmeyer and Crowley.

This outcome measure is considered exploratory. Because of the unanticipated long OS in the control group, initial exploratory subgroup analyses identified that findings in patients enrolled in Russia differed when compared with patients enrolled in the ITT subpopulation and the EU+US subpopulation. For this reason, subjects in Russia were excluded from further exploratory subset analyses that evaluated the effects of genomic parameters known to impact patient responses to anti-EGFR monoclonal antibodies. Removal of the outlier Russian subpopulation of patients provided a patient population that was more homogeneous with respect to their prior treatment regimens, thereby facilitating further exploratory analyses. If a subject had not died, survival time was censored at the last date the subject was known to be alive.

The analysis population was the EU+US DNmCRC analysis set, which is a genomically-defined subpopulation excluding patients with high frequency clonal RAS mutations and BRAF V600E mutations.

OS based on product-limit (Kaplan-Meier) estimates. Confidence intervals for the median are calculated according to Brookmeyer and Crowley.

This outcome measure is considered exploratory. Because of the unanticipated long OS in the control group, initial exploratory subgroup analyses identified that findings in patients enrolled in Russia differed when compared with patients enrolled in the ITT subpopulation and the EU+US subpopulation. For this reason, subjects in Russia were excluded from further exploratory subset analyses that evaluated the effects of genomic parameters known to impact patient responses to anti-EGFR monoclonal antibodies. Removal of the outlier Russian subpopulation of patients provided a patient population that was more homogeneous with respect to their prior treatment regimens, thereby facilitating further exploratory analyses. If a subject had not died, survival time was censored at the last date the subject was known to be alive.

The analysis population was the EU+US TNmCRC analysis set, which is a genomically-defined subpopulation excluding DNmCRC patients with six (6) selected EGFR extracellular domain (ECD) mutations.