Background:
  -  mCRC is incurable and available standard therapies offer a median overall survival
     of approximately 2 years.
  -  Most cases (approximately 95%) of mCRC have an intact expression of DNA mismatch
     repair enzymes (MLH1, MSH2, MSH6, and PMS2), and are commonly classified as having
     mismatch repair proficient (pMMR) or microsatellite stable (MSS) mCRC.
  -  MSS mCRC does not respond to immune checkpoint inhibitor (ICI) therapy whereas
     mismatch repair deficient (or microsatellite instability-high) mCRC is responsive to
     ICI therapy.
  -  Preclinical and clinical studies conducted at the NCI in the Laboratory of Tumor
     Immunology Biology (LTIB) indicate that the combination of programmed cell death
     protein 1 (PD-1) / Programmed death-ligand 1 (PD-L1) blockade, tumor-associated
     antigens (TAA) targeted vaccine, IL15 agonist, and CXCR1/2 inhibition may include
     sufficient immune enhancements to produce anti-tumor activity in MSS mCRC.
  -  Retifanlimab is a humanized IgG4 monoclonal antibody that targets PD-1. Retifanlimab
     has been studied in several clinical trials, and several malignancies, and has a
     safety and clinical activity profile similar to approved anti-PD-1 therapies (e.g.,
     pembrolizumab).
  -  The TriAdeno Vaccine employs 3 adenovirus serotype 5 vectors, encoding three TAAs
     (CEA, MUC1, and brachyury). These vaccines have completed the phase 1 study and are
     safe and well tolerated. Vaccination generates antigen-specific T cell responses to
     CEA, MUC1, and brachyury.
  -  N-803 is an IL-15 agonist that activates and expands T cells and NK cells. N-803
     enhances anti-tumor activity in combination with tumor-targeted vaccines.
     Clinically, multiple studies have demonstrated the safety of tumor-targeted vaccine
     in combination with PD- 1/PD-L1 blockade. Adding N-803 to PD-1/PD-L1 blockade can
     produce antitumor responses in disease states where responses to PD-1/PD-L1 alone
     would not be expected.
  -  SX-682 is a small molecule, orally bioavailable, allosteric antagonist of the
     chemokine receptors CXCR1 and CXCR2. Inhibition of CXCR1 and CXCR2 addresses a major
     component of intratumoral T cell suppression by myeloid-derived suppressor cells and
     tumor-associate macrophages. SX-682 has shown to be tolerable in combination with
     PD- 1/PD-L1 blockade.
Objectives:
  -  Phase I: to describe the safety profile of the Immuno-Oncology (IO) regimens
     consisting of retifanlimab, TriAdeno vaccine, N-803 (A1), and retifanlimab, TriAdeno
     vaccine, N- 803, SX 682 (A2) in participants with metastatic colorectal cancer
     (mCRC).
  -  Phase II: to determine the objective response rate (ORR) (complete response (CR) +
     partial response (PR)) of the IO regimen in mCRC.
Eligibility:
  -  Age >=18 years.
  -  Previously treated metastatic colorectal cancer with measurable disease.
  -  Eastern Cooperative Oncology Group (ECOG) performance status <= 2.
  -  Adequate organ function.
Design:
  -  This is an open-label Phase I/II trial to evaluate the safety and efficacy of the
     Immuno- Oncology regimen, consisting of retifanlimab, TriAdeno vaccine, N-803, and
     SX-682 in participants with mCRC.
  -  During Phase I, we will assess the safety of the three- and the four-drug IO
     regimens.
  -  During Phase II we will continue to evaluate the safety and examine the efficacy of
     the four drug IO regimen. If 0 to 1 of the first 9 participants treated with the
     4-drug IO regimen have a clinical response, defined as complete response (CR) +
     partial response (PR) within the first 24 weeks after treatment initiation, then no
     further participants will be accrued using the quadruple combination. If 2 or more
     of the first 9 participants have a response, then accrual will continue until a
     total of 23 evaluable participants have been treated with a four-drug IO regimen.
  -  Participants will receive treatment in cycles consisting of 28 (+7) days for 2
     years.
  -  To allow for a small number of inevaluable participants, and screen failures the
     accrual ceiling will be set at 60.