Prediction of Invasiveness and Treatment Resistance Behavior of Pituitary Neuroendocrine Tumors Based on Genome-Wide Loss of Heterozygosity

Background:

Pituitary neuroendocrine tumors (PitNETs) are mostly benign, but a small portion exhibit invasive and treatment-resistant behaviors, continuing to grow or metastasize even after surgery, conventional drug treatment, and initial radiation therapy. According to the 2022 European Society of Endocrinology (ESE) clinical practice guidelines, invasive PitNETs are defined as tumors that continue to progress despite standard treatments (surgery, conventional drug treatment, and radiation therapy). Since the WHO removed the classification standard based on indicators like TP53 immunohistochemistry (IHC) overexpression in 2017, this study aims to identify biomarkers that can more accurately predict future treatment resistance of pituitary neuroendocrine tumors.

Research Team and Publication Information:

This study was jointly conducted by Andrew L. Lin, Vasilisa A. Rudneva, Allison L. Richards, and others, involving several authoritative research institutions such as Memorial Sloan Kettering Cancer Center (MSKCC), Boston University, and Columbia University Medical Center. The paper, after revision, was published in April 2024 in the journal “Acta Neuropathologica.”

Research Process:

The study conducted genetic sequencing on two groups of patients: one group of 66 patients who voluntarily participated in genetic sequencing studies before surgery, and another group of 26 patients exhibiting invasive or high-risk characteristics. The study particularly focused on mutation burden and loss of heterozygosity (LOH) phenomena in tumors, with numerous experiments including immunohistochemistry (IHC), whole-exome sequencing (WES), and fluorescence in situ hybridization (FISH).

1. Sample Collection and Processing:

   • Prospective Group: Samples were extracted from the tumors of 66 patients and sequenced.
   • Retrospective Group: Summarized data on 26 confirmed invasive or high-risk patients, particularly noting tumor progression after radiation therapy.

2. Immunohistochemistry Analysis:

   • Utilized various antibodies (e.g., P53, Ki-67) to classify tumor tissues and mismatch repair status.

3. Genomic Sequencing and Data Analysis:

   • Utilized the MSK-IMPACT sequencing platform for next-generation sequencing of tumors, analyzing gene mutations, loss of heterozygosity, and copy number variations.
   • Data were processed and analyzed through various algorithms (e.g., Mutect2, FACETS) to identify mutations and structural rearrangements.

4. Fluorescence In Situ Hybridization (FISH):

   • Verified copy number and heterogeneity of multiple chromosomes in tumors using FISH.

5. Machine Learning Model:

   • Employed a Random Forest algorithm-based classifier to predict the invasiveness and treatment resistance of tumors.
   • The model combined multiple genetic features and clinical data (e.g., loss of heterozygosity, TP53 status) and evaluated the model’s accuracy using the receiver operating characteristic (ROC) curve.

Main Research Results:

1. Gene Mutation Phenomena:

   • Invasive, treatment-resistant PitNETs displayed higher mutation burden and loss of heterozygosity. Particularly in binary classification, TP53 mutations were most common in invasive tumors (12 out of 23 treatment-resistant tumors).

2. Loss of Heterozygosity (LOH) and Its Significance:

   • Invasive PitNETs significantly exhibited widespread chromosomal loss of heterozygosity. Particularly in corticotroph tumors, repeated LOH patterns in 12 specific chromosomes were associated with treatment resistance.
   • FISH confirmed these LOH events were due to the loss of relevant chromosomes or chromosomal regions, indicating chromosomal aneuploidy in invasive PitNETs.

3. Machine Learning Model Prediction:

   • The Random Forest model successfully predicted future treatment resistance of tumors (AUC of 0.83-0.87) by combining genetic and clinical features, where loss of heterozygosity had a higher predictive capability than TP53 mutations.

Research Conclusions and Significance:

The study indicates that loss of heterozygosity (LOH) is a major molecular feature of invasive, treatment-resistant PitNETs, especially in corticotroph tumors, with high predictive value. The newly identified LOH phenomena not only provide new insights into the molecular mechanisms of pituitary neuroendocrine tumors but also help improve early diagnosis and personalized treatment strategies.

Combining machine learning algorithms with genomic analysis offers a novel and effective predictive model for such tumors, underscoring future prospects for early intervention and precision treatment using these molecular biomarkers. Exploiting genomic instability and its driven aberrant signaling pathways may also become new targets for future treatment of invasive PitNETs.

Research Highlights:

1. Loss of Heterozygosity (LOH) as a biomarker for predicting invasiveness and treatment resistance.
2. Discovery of widespread chromosomal LOH events in corticotroph tumors and their early event characteristics.
3. Improvement of prediction accuracy through machine learning models, providing references for clinical treatment.