Баланс РТПХ та ефекту "трансплантат проти лейкемії" — the double-edged sword of donor alloreactivity after allogeneic hematopoietic transplant
After allogeneic hematopoietic stem cell transplantation, donor-derived T-cells are infused (or generated de novo) into a genetically distinct recipient. These T-cells scan recipient tissue via the T-cell receptor (TCR), recognizing peptide-MHC complexes that differ from the donor's own — whether through major HLA mismatch or the more subtle minor histocompatibility antigens (miHAs) present even in fully HLA-matched pairs. This single act of foreign recognition is the common root of both graft-versus-host disease and the graft-versus-leukemia effect: the same alloreactive T-cell clone that attacks a leukemic blast because it presents a mismatched minor antigen can, in principle, attack a healthy epithelial cell presenting that same antigen.
In 1966, Rupert Billingham formalized three requirements for graft-versus-host disease: (1) the graft must contain immunologically competent cells, (2) the recipient must express tissue antigens the graft does not have and therefore perceives as foreign, and (3) the recipient must be incapable of mounting an effective response to eliminate the transplanted cells. These same three conditions are precisely what enable the graft-versus-leukemia effect — a competent donor immune system, recognizably "foreign" antigens on the malignant clone (including tumor-associated and minor histocompatibility antigens), and a host immune system too suppressed by conditioning and malignancy to reject the donor graft.
Minor histocompatibility antigens (miHAs) are polymorphic self-peptides — arising from genetic variation between donor and recipient outside the HLA locus — presented on MHC class I/II. Because many miHAs are expressed broadly across hematopoietic AND non-hematopoietic tissue, a T-cell clone reactive against one miHA can plausibly attack leukemic cells, skin, gut epithelium, and liver simultaneously. This shared antigenic target is the molecular reason the two effects travel together rather than existing as fully independent processes.
Because both effects can be driven by identical alloreactive T-cell clones recognizing identical minor antigens, GVHD and GVL are best understood not as two diseases but as two possible tissue-level outcomes of one shared immunologic event — donor recognition of "non-self."
Not all alloantigens carry equal risk of causing collateral damage. Some minor histocompatibility antigens (e.g., HA-1, HA-2) are expressed almost exclusively on hematopoietic lineage cells — including leukemic blasts — with minimal expression in skin, gut, or liver epithelium. T-cell clones targeting these hematopoietic-restricted antigens can deliver strong antileukemic pressure with comparatively little GVHD.
Other antigens are ubiquitously expressed across essentially all nucleated tissue. Alloreactivity against these broadly expressed targets is far more likely to manifest as classic multi-organ GVHD, because the "foreign" signal is present everywhere donor T-cells patrol — not just at the leukemic site. This distinction — hematopoietic-restricted versus broadly expressed alloantigens — is the conceptual seed for later strategies (Stage 5) that attempt to steer donor immunity toward the former and away from the latter.
The graft-versus-leukemia (GVL) effect is one of the most compelling demonstrations that the immune system itself can function as a durable antineoplastic therapy. Donor T-cells and NK cells patrol the recipient after transplant, recognizing residual malignant cells through alloantigen mismatch, and eliminate them via classical cytotoxic mechanisms — perforin/granzyme release and Fas-FasL–mediated apoptosis. This immune surveillance is a major reason allogeneic transplant achieves lower relapse rates than autologous transplant (which lacks a foreign immune graft) in diseases such as chronic myeloid leukemia, acute myeloid leukemia, and myelodysplastic syndrome.
Three converging lines of clinical evidence established GVL as a real, independent antileukemic mechanism rather than a byproduct of conditioning chemotherapy:
1. Syngeneic (identical twin) transplants show higher relapse rates than allogeneic transplants from HLA-matched but non-identical donors — despite identical conditioning regimens — because syngeneic grafts lack alloreactive potential entirely.
2. T-cell depletion of the graft (removing donor T-cells to reduce GVHD) reproducibly increases relapse risk, sometimes two- to three-fold in some series, directly demonstrating that donor T-cells were suppressing residual disease.
3. Donor lymphocyte infusion (DLI) — reinfusing donor T-cells into a patient who has relapsed after transplant, with no further chemotherapy — can induce durable remission, most dramatically in chronic myeloid leukemia (CML), where DLI alone produces complete cytogenetic remission in a majority of chronic-phase relapses. This is essentially "pure" GVL effect, unclouded by concurrent chemotherapy.
Donor lymphocyte infusion for relapsed chronic-phase CML remains one of medicine's clearest proofs that unmodified donor T-cells, with no additional cytotoxic therapy, can eradicate a hematologic malignancy — establishing cellular alloimmunity as a therapeutic modality in its own right.
The strength of the observed GVL effect varies substantially by disease biology and kinetics:
• Chronic myeloid leukemia (chronic phase): strongest and most reproducible GVL sensitivity — slow tumor kinetics give donor immunity time to act, and DLI alone is often curative for relapse. • Acute myeloid leukemia / myelodysplastic syndrome: meaningful but more variable GVL effect; higher-risk cytogenetics and rapid proliferation can outpace immune control. • Acute lymphoblastic leukemia: GVL effect present but generally weaker and less reliable than in myeloid disease, motivating the newer targeted cellular approaches (e.g., CAR-T) as complements. • Multiple myeloma and lymphoma: GVL contribution documented but heterogeneous across subtypes.
This disease-specific variability is why relapse risk category is a critical variable when weighing the value of tolerating some GVHD against the cost of the associated morbidity — the same degree of GVHD carries different net utility depending on how relapse-prone the underlying disease is.
The identical cytotoxic machinery that eliminates leukemic cells in Stage 2 — donor T-cell recognition of "foreign" antigen followed by perforin/granzyme and Fas-FasL killing — becomes destructive when its target is healthy recipient tissue instead of malignancy. Acute GVHD classically targets skin, gastrointestinal tract, and liver; chronic GVHD can affect nearly any organ system, producing a fibrotic, autoimmune-like syndrome. GVHD remains a leading cause of non-relapse morbidity and mortality after allogeneic transplantation.
Acute GVHD pathogenesis is classically described in three overlapping phases:
Phase 1 — Tissue damage from conditioning: myeloablative chemotherapy and/or radiation damages host epithelium (especially gut mucosa), releasing pro-inflammatory cytokines (TNF-α, IL-1, IL-6) and triggering translocation of microbial products (e.g., LPS) across a compromised gut barrier — a "cytokine storm" priming environment.
Phase 2 — Donor T-cell activation: host antigen-presenting cells (APCs), activated by the inflammatory milieu from Phase 1, present alloantigen (mismatched HLA or minor histocompatibility antigens) to donor T-cells, driving massive clonal expansion and differentiation into effector subsets (Th1, Tc1, and others).
Phase 3 — Effector/target tissue destruction: activated donor cytotoxic T-cells, NK cells, and inflammatory cytokines directly damage target organs — skin (rash, desquamation), gut (diarrhea, mucosal sloughing), and liver (cholestasis, hepatocyte injury) — through the same perforin/granzyme and Fas-FasL pathways that mediate GVL, plus additional cytokine-driven tissue injury (TNF-α, IFN-γ).
Acute GVHD and GVL diverge mainly in target selectivity, not in effector mechanism — the cytokine storm and cytotoxic pathways that destroy gut and skin epithelium in severe GVHD are mechanistically continuous with those that eliminate leukemic blasts.
Chronic GVHD, typically arising after day +100 (though onset timing is now understood as a spectrum rather than a strict cutoff), behaves less like an acute cytotoxic attack and more like an autoimmune, fibrosing disease. Mechanisms include thymic damage impairing central tolerance (allowing autoreactive T-cells to escape negative selection), dysregulated B-cell activation producing pathogenic autoantibodies, and profibrotic cytokine signaling (TGF-β, PDGF) driving collagen deposition.
Chronic GVHD can affect skin (sclerosis), eyes (dry eye, keratoconjunctivitis sicca), mouth (lichenoid changes, sicca), lungs (bronchiolitis obliterans), gut, liver, joints/fascia, and genital tract — producing cumulative disability and markedly reduced quality of life even when not immediately life-threatening. Its severity spectrum, and its own association with GVL protection, is central to the correlation examined in Stage 4.
A consistent finding across large transplant registries: patients who develop mild-to-moderate GVHD tend to have lower subsequent relapse rates than patients with no GVHD at all — while patients with severe GVHD do not necessarily gain additional protection and instead accumulate disproportionate morbidity and non-relapse mortality. This produces a characteristic non-linear relationship, sometimes described loosely as a "sweet spot," but it is critical to interpret this correlation cautiously rather than as license to induce GVHD deliberately.
The observation that "some GVHD associates with less relapse" is a population-level statistical pattern, not a prescription. Several confounders complicate direct causal interpretation:
• Immunosuppression asymmetry: patients who never develop GVHD are, almost by definition, more immunosuppressed or have less alloreactive grafts — the same immunologic quiescence that spares them GVHD may also reduce their GVL protection, meaning weak alloreactivity — not absence of GVHD per se — may be the common cause of both outcomes. • Survivorship and timing biases: patients must survive long enough to relapse or to be scored for GVHD, and GVHD-free early deaths from other causes can distort simple incidence comparisons if not handled with proper time-dependent statistical methods (e.g., landmark analysis, competing risk models). • Severity, not mere presence: the protective association is generally strongest for mild-to-moderate GVHD; once GVHD becomes severe, additional relapse protection is limited while treatment-related mortality rises sharply — the relationship is not simply "more GVHD, less relapse."
The clinically responsible reading of this correlation is: mild GVHD is a marker of a functioning, alloreactive graft-versus-tumor immune system — not a goal to be pursued for its own sake. Deliberately inducing or undertreating GVHD to "boost" GVL is not supported as a treatment strategy; the priority remains preventing severe GVHD while preserving immune surveillance.
This correlation informs — but does not dictate — several real clinical practices:
• Calibrated immunosuppression tapering: in patients at high relapse risk, clinicians may taper GVHD prophylaxis somewhat more cautiously or monitor more closely for early relapse signals, rather than maximizing immunosuppression to eliminate any GVHD risk. • Preemptive/pre-relapse donor lymphocyte infusion: for high-risk disease (e.g., detectable minimal residual disease, high-risk cytogenetics), clinicians may consider early DLI or immunosuppression withdrawal to harness GVL — accepting some GVHD risk in a context where relapse risk clearly outweighs it. • Risk-adapted prophylaxis intensity: lower relapse-risk patients (e.g., favorable cytogenetics, deep remission at transplant) are generally managed toward more aggressive GVHD prevention, since the marginal GVL benefit of tolerating GVHD matters less when baseline relapse risk is already low.
This is precisely the logic captured by the balance assessment in this simulator: the same GVHD severity can be net favorable or net unfavorable depending on the relapse-risk context of the underlying disease.
If GVHD and GVL share a common mechanism but differ in target selectivity, the central research question becomes: can donor immunity be steered — pharmacologically, genetically, or through graft engineering — toward leukemic targets and away from healthy tissue? This is an active, still-incomplete area of transplant immunology, with several complementary strategies each addressing a different piece of the separation problem.
Rather than removing all donor T-cells (which sacrifices GVL, Stage 2) or leaving the graft unmodified (which risks severe GVHD, Stage 3), selective graft engineering aims for a middle path:
• Naïve T-cell depletion: memory T-cells are thought to contribute disproportionately to early GVL activity with somewhat less GVHD-driving potential than naïve subsets in some models; grafts depleted of naïve T-cells while retaining memory populations are an active investigational approach. • Regulatory T-cell (Treg) enrichment or co-infusion: Tregs can suppress alloreactive effector T-cell expansion broadly, dampening GVHD, while still permitting substantial antileukemic activity in early studies — the challenge is dosing and timing to avoid blunting GVL along with GVHD. • Photodepletion and alloanergization: ex vivo techniques that selectively eliminate or functionally inactivate the specific donor T-cell clones that react most strongly against recipient antigens, while sparing the broader T-cell repertoire needed for pathogen and tumor surveillance.
A conceptually different strategy sidesteps broad donor T-cell alloreactivity altogether by engineering antigen-specific effector cells:
• CAR-T and TCR-engineered T-cells directed at leukemia-associated antigens (e.g., CD19 in B-lineage ALL, CD33/CLL-1 in AML) can deliver potent antileukemic killing without relying on unpredictable, tissue-wide alloreactivity — in principle decoupling antileukemic potency from GVHD risk, since the receptor specificity is defined rather than emergent. • NK cell-based approaches, particularly from KIR-mismatched donors, can exploit "missing-self" recognition to kill leukemic cells with comparatively limited GVHD-inducing potential, since NK alloreactivity operates through different receptor biology than T-cell TCR-mediated GVHD. • Post-transplant cyclophosphamide (PTCy) selectively eliminates rapidly proliferating alloreactive T-cell clones shortly after transplant while sparing quiescent T-cells (including regulatory populations) — a pharmacologic timing strategy that has meaningfully reduced severe GVHD in haploidentical transplant without eliminating GVL activity.
None of these strategies yet achieves complete separation of GVL from GVHD — the two effects remain mechanistically entangled at their root. The realistic near-term goal is incremental improvement in the ratio of benefit to harm, not a perfect decoupling, making thoughtful risk-adapted clinical judgment (as explored across all five stages here) still essential.