GUT: Liang Tingbo's team from the First Hospital of Zhejiang University developed an early screening model for pancreatic cancer and liver cancer, and the accuracy rate exceeded 90%!

The incidence and mortality of malignant tumors of the digestive system remain high, especially liver cancer and pancreatic cancer, which have the characteristics of insidious onset and rapid progression, and most advanced liver cancer and pancreatic cancer have lost the opportunity for surgery, resulting in a very poor prognosis of patients [1].

Therefore, early screening is essential
for liver and pancreatic cancer.

At present, the serum markers of liver cancer and pancreatic cancer used in clinical practice are alpha-fetoprotein (AFP) and cancer antigen 19-9 (CA19-9), respectively, but the sensitivity and specificity of these two indicators do not meet the needs of early screening [2,3].

In recent years, liquid biopsy technology based on indicators such as circulating tumor DNA and exosomes in peripheral blood has been successfully used for cancer detection, but there is still no indicator that can be applied to early clinical screening [4].

So the scientific community thought of detecting the immune system response of tumor patients, and a large number of studies have shown that changes in the number and composition of immune cells occur during tumorigenesis and progression, and this change is reflected in systemic immune disorders and changes in peripheral blood immune cell subsets [4,5].

Therefore, detecting the composition and number of immune cells through peripheral blood may become a new method
of early cancer screening.

Recently, the research team led by Professor Liang Tingbo of the First Affiliated Hospital of Zhejiang University School of Medicine published an important research result
in the top international journal of gastroenterology, Gut 。 This study detected the composition and number of peripheral blood immune cells in patients with liver cancer, pancreatic cancer and healthy people by mass spectrometry flow cytometry (CyTOF), determined the changes in specific immune cell subsets in liver cancer and pancreatic cancer, and constructed an early diagnosis model of liver cancer and pancreatic cancer based on this, with sensitivity and specificity of more than 90%, with high diagnostic efficacy [7], thus providing a new strategy
for the early diagnosis and screening of liver cancer and pancreatic cancer.

Screenshot of the article cover

In order to obtain accurate and universal conclusions, the researchers collected samples from 2348 subjects from 15 centers, including 633 healthy controls, 1131 patients with liver disease (341 cases of benign liver disease and 790 cases of liver cancer), 584 patients with pancreatic disease (208 cases of benign pancreatic disease and 376 cases of pancreatic cancer), and after completing sample quality control, all samples were tested for CyTOF to to identify 32 major immune cells (see Figure 1).
The difference in
the composition and number of these immune cells between cancer patients and healthy people or patients with benign diseases was analyzed.

Figure 1.
Identification of immune cells in peripheral blood

The researchers found that compared with healthy people and patients with benign diseases, there are obvious differences in the subset of peripheral blood immune cells in patients with liver cancer and pancreatic cancer: the proportion of naïve CD4 T cells, naïve CD8 T cells and effector CD8 T cells in liver cancer patients decreases, while the proportion of plasma cells and central memory CD4 T cell monocytes increases, and some subpopulations represented by NK cells increase in proportion with the progression of tumor stage, and these phenomena also exist in pancreatic cancer patient samples
。

That is to say, in the peripheral blood of patients with malignant tumors, there is indeed a different subset
of immune cells than in healthy people and patients with benign diseases.
In order to rationally construct the prediction model, the researchers divided the subjects into three groups: model construction group, internal verification group and external verification group, and identified 7 markers and 16 cell subsets for the diagnosis of liver cancer and 8 markers and 11 cell subsets for the diagnosis of pancreatic cancer through random forest algorithm, and based on this, the model was constructed, and the diagnostic index
of peripheral blood immune score (PBIScore) was obtained.

PBIScore reflects the change in the proportion of tumor-related immune cell subsets in human peripheral blood, and is similar to the commonly used tumor markers AFP and CA19-9, PBIScore is also a non-invasive blood test
.
So compared with the "front wave", can the "back wave" of PBIScore be worthy of the great responsibility?

The researchers found that the AUC values (response diagnostic efficacy) of PBIScore in the liver cancer diagnostic model group, internal verification group and external verification group reached 0.
98, 0.
91 and 0.
85, respectively, and the AUC values in the pancreatic cancer diagnostic model group, internal verification group and external verification group were 0.
98, 0.
89 and 0.
89
, respectively.
In other words, the AUC value of PBIScore is higher than the two traditional indicators of AFP and CA19-9, but the advantage is not very significant
.

In order to further enhance the diagnostic performance of PBIScore, the researchers combined it with AFP or CA19-9 to construct a new scoring model - iPBIscore, which further enhanced its diagnostic performance: the AUC values of iPBIscore in the liver cancer diagnostic model group, internal verification group and external verification group were as high as 0.
99, 0.
97 and 0.
96, respectively, and the AUC value of pancreatic cancer diagnosis reached 0.
99.
Both 0.
98 and 0.
97 are significantly higher than PBIScore or AFP/CA19-9 alone (Figure 2).

Figure 2.
Efficacy of diagnostic models for liver and pancreatic cancer based on PBIScore

Most markers can play a good role in the diagnosis of advanced or advanced tumors, but when it comes to early tumor diagnosis, early diagnosis is the most important
thing.
Therefore, the researchers tested the effect
of this PBIScore-based model on the diagnosis of early-stage liver cancer and pancreatic cancer.

The results showed that the AUC values of AFP/CA19-9, PBIScore model and iPBIScore model for the diagnosis of early liver cancer were 0.
81, 0.
90 and 0.
96, respectively, and the AUC values for the diagnosis of early pancreatic cancer were 0.
88, 0.
89 and 0.
95, respectively (Figure 3).

It can be seen that the PBIScore model has strong diagnostic ability for early pancreatic cancer and liver cancer, and has a good complementary effect on traditional tumor markers, and the combination of the two can greatly improve the early diagnosis rate
of liver cancer and pancreatic cancer.

Figure 3.
Diagnostic effect of PBIScore-based model on early liver cancer and pancreatic cancer

Based on the good diagnostic effect of the PBIScore model, the researchers speculate that the immune cell subsets included in the model are potentially related
to pathological features such as tumor markers, differentiation degree and stage.

It was verified that the proportion of peripheral blood mononuclear cells and central memory CD8 T cells in AFP-positive patients was significantly lower than that in AFP-negative patients, and the proportion of centrally differentiated and high-differentiated HCC patients was much higher than that in poorly differentiated patients (Figure 4A).

In addition, peripheral blood immune cell subsets were significantly correlated with pancreatic cancer tumor location and disease stage (Figure 4B).

Therefore, the imbalance of these peripheral immune cell subsets has an important impact on
tumorigenesis and progression.

Figure 4.
Correlation between immune cell subsets and pathological features of pancreatic and liver cancer in the PBIScore model

In conclusion, this study revealed the influence of peripheral blood immune cell subset disorders on tumorigenesis and development through multi-center, large-sample peripheral blood immune cell subset detection, developed an early screening model that can efficiently diagnose liver cancer and pancreatic cancer, and proposed a new strategy and direction
for early cancer screening.

References:

[1] Sharma P, Hassan C.
Artificial Intelligence and Deep Learning for Upper Gastrointestinal Neoplasia.
Gastroenterology.
2022 Apr; 162(4):1056-1066.
doi: 10.
1053/j.
gastro.
2021.
11.
040.

[2] Melbye M, Wohlfahrt J, Lei U, et al.
Alpha-Fetoprotein levels in maternal serum during pregnancy and maternal breast cancer incidence.
J Natl Cancer Inst 2000; 92:1001–5.
doi:10.
1093/jnci/92.
12.
1001

[3] Ballehaninna UK, Chamberlain RS.
The clinical utility of serum CA 19-9 in the diagnosis, prognosis and management of pancreatic adenocarcinoma: An evidence based appraisal.
J Gastrointest Oncol.
2012 Jun; 3(2):105-19.
doi: 10.
3978/j.
issn.
2078-6891.
2011.
021.

[4] Allen BM, Hiam KJ, Burnett CE, et al.
Systemic dysfunction and plasticity of the immune macroenvironment in cancer models.
Nat Med 2020; 26:1125–34.
doi:10.
1038/s41591-020-0892-6.

[5] Wang L, Simons DL, Lu X, et al.
Connecting blood and intratumoral Treg cell activity in predicting future relapse in breast cancer.
Nat Immunol 2019; 20:1220–30.
doi:10.
1038/s41590-019-0429-7.

[6] Zhang Q, Mao Y, Lin C, et al.
Mass cytometry-based peripheral blood analysis as a novel tool for early detection of solid tumours: a multicentre study.
Gut.
2022 Sep 16:gutjnl-2022-327496.
doi: 10.
1136/gutjnl-2022-327496.