A multicenter prospective validation cohort does not confirm the diagnostic yield of FDG PET/CT imaging in kidney allograft subclinical rejection

This multicenter prospective study of 185 kidney transplant recipients found that [18F]FDG PET/CT imaging, measured by mean standardized uptake value ratios, fails to reliably detect or rule out subclinical acute rejection in stable patients at three months post-transplantation.

The Gist

Imagine you have just bought a brand-new, high-performance car (a kidney transplant). You want to make sure the engine is running perfectly, but you can't just pop the hood and look inside without potentially damaging the delicate parts. In the medical world, the "gold standard" for checking a kidney transplant is a biopsy: a doctor uses a needle to take a tiny sample of the kidney tissue to look at it under a microscope. It's like taking a tiny scoop of cake to taste it and see if the ingredients are right.

However, taking that scoop is invasive. It hurts a little, carries a small risk of bleeding, and you can't do it every week. So, doctors have been hunting for a "magic scanner" that could look at the kidney from the outside and say, "Everything is fine, no need to poke it!"

The Proposed Magic Scanner: [18F]FDG PET/CT

The researchers in this study tested a specific type of scanner called a PET/CT scan.

Think of your immune system as a team of security guards. When a kidney transplant is being rejected (even if the patient feels perfectly fine and has no symptoms), these security guards start swarming the kidney, trying to attack it. These guards are very active and hungry; they burn a lot of energy.

The PET scan uses a special radioactive sugar called [18F]FDG. When you inject this sugar into the patient, the hungry security guards (inflammatory cells) gobble it up. The scanner then takes a picture, lighting up the areas where the guards are eating.

• Bright spots = Lots of guards eating = Possible rejection.
• Dim spots = No guards eating = Everything is calm.

The Previous Theory

In a smaller, earlier study, these same researchers thought they found a "magic number." They believed that if the brightness of the kidney (compared to a muscle in the back called the psoas) was below a certain level (2.4), you could be 98% sure the kidney was safe. They thought this scanner could replace the needle biopsy for many patients.

The Big Test: The Multicenter Validation

This new paper is the "final exam." The researchers didn't just test this on 50 people at one hospital; they tested it on 185 patients across four different hospitals in Belgium. They used different types of scanners (some older, some newer digital ones) to make sure the results were real and not just a fluke of one specific machine.

They did two things for every patient:

1. Gave them the PET scan.
2. Immediately did the biopsy (the needle poke) to see what was actually happening inside.

The Results: The Magic Number Didn't Work

Here is the disappointing news for the "magic scanner" idea: It didn't work as hoped.

• The False Negatives: Several patients had a "quiet" scan (a low number below the 2.4 threshold), but when the doctors looked at the biopsy, they found rejection happening anyway. The scanner said, "All clear!" but the biopsy said, "There's a problem!"
• The Overlap: The brightness levels of healthy kidneys, kidneys with minor issues, and kidneys with rejection were all jumbled together. You couldn't draw a clear line to separate them.
• The Conclusion: The study found that you cannot rely on this PET scan to rule out rejection. If you used the scan to skip the biopsy, you might miss patients who actually need treatment.

A Simple Analogy

Imagine you are trying to detect if a house has a fire.

• The Biopsy is like sending a firefighter inside to check every room. It's thorough but annoying and risky.
• The PET Scan is like looking at the house from the street to see if the windows are glowing red.

In this study, the researchers hoped that if the windows weren't glowing red, the house was definitely safe. But they found that some houses with small fires inside had windows that looked perfectly normal from the street. If you had trusted the windows, you would have missed the fires.

The Takeaway

While the idea of a non-invasive, painless scan to replace needle biopsies is wonderful, this study proves that this specific scan is not ready for prime time to replace the biopsy for kidney transplant patients.

The doctors still need to do the needle biopsy to be sure. However, the researchers suggest that maybe we need to look at the scan pictures in a smarter way (using AI or "radiomics" to analyze the texture, not just the brightness) to see if we can find a better pattern in the future. For now, the needle remains the most reliable tool.

Technical Summary

Title

A multicenter prospective validation cohort does not confirm the diagnostic yield of [18F]FDG PET/CT imaging in kidney allograft subclinical rejection.

1. Problem Statement

• Clinical Context: Subclinical Rejection (SCR) in kidney transplant recipients (KTRs) is defined as histological evidence of acute rejection in a clinically stable patient. Early detection and treatment of SCR are crucial to prevent late graft failure and fibrosis.
• Current Standard: The gold standard for diagnosing SCR is a surveillance allograft biopsy. However, this is an invasive procedure with a ~7% complication rate, risks sampling errors, and suffers from inter-observer variability.
• Proposed Solution: Non-invasive imaging using [18F]FDG PET/CT was hypothesized to serve as a "rule-out" test. The premise is that inflammatory cells recruited during rejection exhibit increased glucose metabolism, detectable via [18F]FDG uptake.
• The Gap: A previous monocentric pilot study suggested that a kidney-to-psoas muscle mean Standardized Uptake Value ratio (mSUVR) < 2.4 could exclude SCR with 98% negative predictive value. This study aims to validate that specific threshold in a larger, multicenter setting.

2. Methodology

• Study Design: Prospective, multicenter validation cohort study involving 185 adult kidney transplant recipients.
• Setting & Timeline: Conducted from January 2021 to March 2025 across three independent centers in Belgium (UZA, UZB, CHU Liège).
• Protocol:
  • Patients underwent a surveillance transplant biopsy at approximately 3 months (median 91 days) post-transplantation.
  • [18F]FDG PET/CT scans were performed within 24 hours of the biopsy (median 0.51 days) on stable patients, prior to any immunosuppression changes.
  • Exclusions: Patients with BK polyomavirus viremia (>3 log10 copies) or uninterpretable histology were excluded.
• Imaging Quantification:
  • Mean Standardized Uptake Value (mSUV) was measured in the kidney cortex using 4 volumes of interest (VOIs).
  • Normalization: The kidney mSUV was normalized to the psoas muscle mSUV to create the mSUVR (Target-to-Background ratio), mitigating device-specific sensitivity disparities.
  • Equipment: Various PET/CT systems were used (analog and digital, including Philips, Siemens, and GE), necessitating stratification by center in statistical analysis.
• Histopathology:
  • Biopsies were assessed by blinded pathologists using the Banff 2022 classification.
  • Groups: Patients were categorized as Normal, Borderline, or Subclinical Rejection (SCR). SCR included T-cell-mediated rejection (TCMR) and DSA-negative/C4d-negative microvascular inflammation (MVI).
• Statistical Analysis:
  • Correlation between mSUVR and Banff "ti" (total inflammation) scores.
  • Multivariate logistic regression (stratified by center) to assess the association between mSUVR and the risk of non-normal histology or biopsy-proven SCR.
  • Evaluation of the previously proposed 2.4 mSUVR threshold.

3. Key Contributions

• First Multicenter Validation: This is the first prospective, multicenter study attempting to validate the [18F]FDG PET/CT mSUVR threshold for SCR exclusion in kidney transplants.
• Rigorous Protocol: The study employed a standardized protocol across different imaging modalities (analog vs. digital PET/CT) and utilized the latest Banff 2022 classification criteria.
• Negative Validation: It provides high-quality evidence refuting the utility of the specific mSUVR < 2.4 threshold as a reliable rule-out test for SCR in the early post-transplant period.

4. Key Results

• Cohort Demographics: 185 patients (mean age 55.0 years).
• Histological Distribution:
  • Normal: 158 patients (85.4%)
  • Borderline: 18 patients (9.7%)
  • SCR: 9 patients (4.9%) (6 TCMR, 3 MVI)
• Correlation Analysis:
  • No significant correlation was found between mSUVR and the Banff "ti" score (R=0.032, p=0.67) or the composite acute Banff score.
• mSUVR Values by Group:
  • Normal: 2.33 [1.97–2.93]
  • Borderline: 2.71 [2.50–3.33]
  • SCR: 2.42 [2.27–3.14]
  • Note: The SCR group's median mSUVR (2.42) was not significantly higher than the Normal group, and the ranges heavily overlapped.
• Threshold Performance (mSUVR > 2.4):
  • Only 5 out of 9 SCR cases (55.6%) exceeded the 2.4 threshold.
  • 87 out of 176 non-SCR cases (49.4%) also exceeded the threshold.
  • The difference was not statistically significant (p=0.75).
• Multivariate Analysis:
  • In models stratified by center, mSUVR was not significantly associated with the risk of non-normal histology (OR=4.11, p=0.07) or biopsy-proven SCR.
  • Donor age was the only significant predictor for non-normal histology (OR=1.05, p=0.02).

5. Significance and Conclusion

• Clinical Impact: The study definitively concludes that [18F]FDG PET/CT using the mSUVR metric cannot reliably rule out subclinical rejection in stable kidney transplant recipients at 3 months post-transplant.
• Implication for Practice: The previously proposed threshold of mSUVR < 2.4 should not be used to avoid surveillance biopsies, as it fails to detect a significant proportion of rejection cases (high false-negative rate) and yields many false positives.
• Future Directions: The authors suggest that while [18F]FDG PET/CT alone is insufficient, future research should explore:
  • Radiomics: High-throughput extraction of quantitative features from images to better characterize tissue phenotypes.
  • Alternative Biomarkers: Combining imaging with other non-invasive modalities (e.g., gene expression, omics) to improve diagnostic yield.
  • Refining the "Borderline" Phenotype: Better distinguishing between normal, borderline, and rejection states using advanced imaging or molecular markers.

In summary, this study serves as a critical "reality check" for the field, demonstrating that the metabolic activity of inflammatory cells in early subclinical rejection is not distinct enough from normal physiological uptake to serve as a standalone diagnostic rule-out tool via [18F]FDG PET/CT.