Patient-derived xenograft implantation success in immunodeficient mice
Patient-derived xenograft (PDX) models begin with a race against ischemic time. A surgical or biopsy tumor specimen must be transported from the operating room to the animal facility and implanted while cell viability remains high — every hour of delay measurably lowers the odds of successful engraftment.
Traditional cancer cell lines are grown in plastic dishes for years or decades, during which they undergo genetic drift, lose stromal interactions, and adapt to 2D culture conditions far removed from a real tumor. PDX models instead implant a piece of an actual patient tumor directly into a living host, preserving the three-dimensional architecture, clonal heterogeneity, and — for the first several passages — much of the original stromal composition.
Because PDX tumors retain the mutational landscape and drug-response phenotype of the donor tumor far better than cell-line xenografts, they have become a standard tool in preclinical oncology for testing candidate therapies, predicting patient response (co-clinical trials), and building "living biobanks" that mirror the genomic diversity of human cancer.
Studies comparing PDX drug response to the originating patient's clinical response report concordance rates of 85–90% for several solid tumor types — substantially higher than cell-line xenograft models.
Specimen handling protocol typically follows these steps:
• Surgical or core-needle biopsy tissue is placed immediately into cold (2–8°C) serum-free transport media (e.g., DMEM/F12 with antibiotics) — never frozen, never in formalin • A pathologist confirms viable tumor content and grade from an adjacent tissue section, ensuring the implanted piece is representative and not purely necrotic tissue • The specimen is transported to the vivarium under a validated chain-of-custody, ideally within 1–2 hours and no later than 4 hours • Under a biosafety cabinet, necrotic and connective tissue is trimmed away, and the viable tumor core is retained for fragmentation
Every step is logged for the biobank record — donor demographics, tumor site, stage, prior treatment history, and histopathology — because these variables will later determine which PDX lines are useful for specific research questions.
PDX generation requires informed patient consent under an IRB-approved protocol, typically obtained at the time of surgical consent. Tissue not needed for diagnostic pathology is redirected to the research biobank. All animal procedures require IACUC approval and adherence to strain-appropriate housing (individually ventilated cages, autoclaved bedding, sterile diet) since the recipient mice are severely immunodeficient and vulnerable to opportunistic infection.
Large academic and commercial biobanks (e.g., NCI PDXNet, Jackson Laboratory PDX, Champions Oncology) now maintain thousands of annotated, cryopreserved PDX lines spanning dozens of cancer types, distributed to researchers worldwide for comparative and mechanistic studies.
The trimmed tumor specimen is cut into small, uniform fragments and surgically placed under the skin of an immunodeficient mouse. The choice of recipient strain and implantation technique are the two variables with the greatest influence on whether the tumor will "take."
A human tumor fragment carries a full complement of human antigens that a normal mouse immune system would reject within days. PDX models therefore require mice with progressively deeper immune deficits:
• Nude (nu/nu): lack a thymus, so no mature T cells — but retain functional NK cells and some innate immunity, giving moderate take rates for immunogenic tumors • SCID: additionally lack functional B cells (no adaptive antibody response) due to a DNA repair defect, but retain NK cell activity • NOD-SCID: combines SCID with NOD background defects in innate immunity (reduced complement, dysfunctional macrophages/dendritic cells), improving take rates further • NSG / NOG (NOD-scid IL2rγ-null): the current gold standard — the IL2 receptor gamma chain knockout ablates NK cell development entirely, leaving essentially no adaptive or innate lymphoid rejection machinery
The deeper the immunodeficiency, the higher the engraftment rate across tumor types, but also the greater the husbandry burden — NSG mice must be kept in strict specific-pathogen-free conditions because they cannot fight off infections that a normal mouse would clear easily.
Take rates for the same tumor specimen can differ dramatically by strain: a poorly immunogenic breast cancer fragment might show ~20% take in nude mice versus >70% in NSG mice, purely due to residual NK-cell-mediated rejection in the less deficient strain.
Under general anesthesia, a small (5–8 mm) skin incision is made on the dorsal flank, and a subcutaneous pocket is bluntly dissected. One tumor fragment is inserted per pocket, and the incision is closed with a wound clip or suture. Some protocols instead implant into the mammary fat pad (for breast cancer) or renal capsule for maximal engraftment efficiency, but flank subcutaneous implantation remains the workhorse technique due to its speed, low morbidity, and ease of caliper-based tumor monitoring.
Surgical technique matters: crushing the fragment, over-handling the tissue, or allowing it to dry out during the procedure all measurably reduce viability. Facilities that run high volumes standardize the entire process to under 5 minutes of ex vivo time per fragment, often implanting several fragments from the same donor into multiple mice in parallel to hedge against engraftment failure.
Many protocols co-inject or coat the fragment with growth-factor-reduced Matrigel (a basement-membrane extract rich in laminin and collagen IV) before implantation. This scaffold supports early cell survival and can modestly improve take rates for fragile or poorly vascularized tumor types by providing a provisional extracellular matrix while the host begins to build new blood vessels into the graft.
Estrogen pellet supplementation is standard practice for hormone-receptor-positive breast cancer PDX models, since female mice do not otherwise produce enough circulating estrogen to support tumor growth long-term.
The first two to six weeks after implantation are the make-or-break period for a PDX fragment. Cells at the fragment core survive only by diffusion of oxygen and nutrients until new mouse-derived blood vessels grow in — a race that many fragments lose.
Immediately after implantation, the tumor fragment has no blood supply of its own — it survives purely on passive diffusion of oxygen and nutrients from surrounding host tissue, which is only effective up to roughly 150–200 micrometers from the nearest vessel. Cells beyond this radius become progressively hypoxic, stabilizing HIF-1α and triggering secretion of angiogenic factors (VEGF, bFGF, PDGF) that recruit host endothelial cells to sprout new capillaries into the graft.
This "angiogenic switch" typically begins within the first 7–10 days but takes several weeks to establish a mature, perfusing microvascular network. Fragments that fail to trigger sufficient angiogenesis — often due to low intrinsic angiogenic potential, poor initial cell viability, or unfavorable implantation site — undergo progressive central necrosis and are eventually resorbed by host macrophages, appearing clinically as a shrinking, then disappearing, nodule.
Because the new vasculature is host (mouse)-derived rather than the original human tumor vessels, from this point onward the PDX model is a genuine chimera: human tumor parenchyma growing inside mouse-derived stroma and blood supply.
During the engraftment window, mice are palpated and their flanks measured with digital calipers 1–2 times per week. Most PDX programs define "engraftment" operationally as a tumor reaching 100–150 mm³ in volume (calculated as length × width² / 2), which typically requires the fragment to have first survived the hypoxic bottleneck and then transitioned into detectable exponential growth.
Mice that show no palpable mass by 8–12 weeks are usually censored as engraftment failures. Because the outcome is essentially binary per mouse (engraft or fail), programs implant multiple fragments from the same donor across several mice to increase the chance that at least one successfully establishes — a successful line can then be expanded by re-implanting fragments from that first generation (F1) tumor into additional mice.
Retrospective studies across large PDX cohorts consistently identify several predictors of engraftment success:
• Tumor grade/aggressiveness: poorly differentiated, high-grade tumors engraft more readily than well-differentiated, indolent ones • Tumor type: melanoma, pancreatic, and triple-negative breast cancers show consistently high take rates (60–90%); well-differentiated luminal breast and low-grade prostate cancers are notoriously difficult (<20%) • Prior treatment: heavily pretreated, chemo-resistant tumors often engraft at higher rates than treatment-naive tumors, likely reflecting selection for aggressive clones • Fragment viability and handling speed: fresher, gently handled tissue with minimal ischemic time consistently outperforms delayed or rough-handled specimens • Host strain immunodeficiency depth: as covered in Stage 2, deeper immune deficiency raises take rates across virtually all tumor types
Once a fragment survives the engraftment bottleneck, growth typically accelerates into a reproducible exponential phase. This is the point at which a PDX line becomes a usable research tool — expandable, cryopreservable, and ready for drug efficacy studies.
Once sufficiently vascularized, the tumor transitions from a slow, diffusion-limited growth regime to a much faster exponential phase, similar in shape to the growth curve of a cell-line xenograft but usually with a longer initial lag. Growth rate at this stage is influenced by the tumor's intrinsic proliferative index, the density of the new host vasculature, and continued replacement of human stromal elements (fibroblasts, immune infiltrate) with mouse-derived counterparts over successive passages.
Once a first-generation (F1) tumor reaches roughly 1000 mm³, it is harvested, re-fragmented, and re-implanted into a larger cohort of naive mice (F2) — the first true expansion step that converts a single successful engraftment into a scalable model.
A portion of every successfully expanded PDX line is cryopreserved in liquid nitrogen (typically in DMSO-containing freeze medium) at each passage, creating a permanent backup independent of any single living mouse colony. This is essential both for biobanking (so a line is never lost to a failed cage or contamination event) and for distributing models to collaborators, who can thaw and re-establish the line in their own facility.
Each banked vial is annotated with passage number, freeze date, and viability on post-thaw re-implantation testing — typically a subset of banked lines are periodically test-thawed to confirm they still engraft reliably before being released for distribution.
Once a PDX line is established and expandable, tumor-bearing mice are enrolled into drug efficacy studies when tumors reach a defined size window (commonly 150–250 mm³), at which point they are randomized across treatment arms to balance starting tumor volume. Studies then track tumor volume over time by caliper measurement, alongside body weight to monitor drug tolerability, generating growth curves used to calculate tumor growth inhibition (TGI%) relative to vehicle control.
Because PDX tumors retain much of the original patient's genomic and histological features, these efficacy readouts are considered more clinically predictive than equivalent studies in established cell-line xenografts, particularly for targeted therapies matched to specific driver mutations present in the donor tumor.
No single implantation predicts how a tumor type will behave — take-rate statistics only become meaningful when aggregated across hundreds of patient samples. These pooled numbers guide which cancer types are realistic candidates for PDX-based research programs.
Pooled data across large PDX programs (NCI PDXNet, EurOPDX, Jackson Laboratory) consistently shows take rates ranging from under 10% to over 90% depending on tumor histology. Aggressive, high-grade, treatment-resistant cancers such as triple-negative breast cancer, pancreatic ductal adenocarcinoma, and melanoma engraft readily, likely because the biological features that make these tumors clinically aggressive — high proliferative index, angiogenic capacity, invasive stroma-remodeling behavior — are the same features that let them establish quickly in a foreign host.
Conversely, indolent, well-differentiated, hormone-driven cancers like ER-positive luminal breast cancer or low-grade prostate adenocarcinoma engraft poorly, in part because they depend on a specific hormonal and stromal microenvironment that a mouse host only partially replicates, and in part because their low proliferative index gives the initial hypoxic bottleneck more time to cause fragment failure.
Because take rate correlates with aggressiveness, PDX biobanks are inherently biased toward representing more aggressive disease — a limitation researchers must account for when using take-rate cohorts to model the full clinical spectrum of a cancer type.
PDX programs typically report take rate as (mice with confirmed tumor growth) / (mice implanted), often with 95% confidence intervals given the modest sample sizes typical per patient sample (commonly 4–10 mice implanted per donor). Time-to-engraftment is also tracked as a Kaplan-Meier-style curve, since censored mice (no growth observed at study end) must be handled statistically rather than simply excluded.
Regression models incorporating tumor grade, stage, prior treatment line, patient age, and mouse strain can predict engraftment probability for a new sample with reasonable accuracy, allowing programs to triage limited biobanking resources toward samples most likely to successfully establish.
For a research program deciding whether PDX modeling is feasible for a given cancer type, historical take-rate statistics are the single most important planning input: a program targeting ER+ breast cancer must budget for implanting many more donor samples per usable model than one targeting melanoma, and must plan for a longer, less predictable timeline to a usable cohort.
Many centers now supplement low-take histologies with humanized mouse hosts (see the companion "Humanized Mouse Immune System Reconstruction" simulation) or orthotopic implantation at the tissue-matched anatomical site (see "Orthotopic vs Subcutaneous Implantation"), both of which can meaningfully raise take rates for otherwise difficult tumor types by better replicating the native tumor microenvironment.