The accuracy of blood test for the three major cancers is higher than 90%!

Due to the aging of the population and changes in other risk factors, the incidence and mortality of cancer are increasing worldwide
.

Among them, prostate cancer is the second most common cancer in men , and breast cancer is one of the main causes of cancer-related deaths in women.

, Breast cancer

Cancer screening methods are designed to detect cancer early in high-risk groups
.

At present, people pay more attention to the development of cancer screening methods based on the examination of peripheral blood (including liquid biopsy), such as circulating cells in the blood, platelets, extracellular vesicles, mRNA, miRNA, proteins, cell-free DNA (cfDNA) and blood In circulating tumor DNA (ctDNA), metabolism has also attracted the attention of cancer screening research

The plasma lipids of patients with kidney cancer, breast cancer and prostate cancer were quantitatively determined and compared with those of healthy controls

Heparin plasma samples were obtained from 289 patients with cancer (breast cancer, prostate cancer, or kidney cancer) and 192 volunteers with no history of malignant disease
.

The sample set is divided into training and verification groups.

The sample order is randomized for extraction and UHPSFC/MS measurement to rule out any possible deviations
.

Liposome analysis was performed on human plasma during the discovery phase.

The sample order is randomized for extraction and UHPSFC/MS measurement to rule out any possible deviations

Figure 3 summarizes the specificity and sensitivity values ​​of different cancer stages and the accuracy values ​​of different cancer types

UHPSFC/MS measurements are repeated several months later to verify the repeatability of the results.
These repeated measurements are called the confirmation phase
.

A total of 138 lipids were quantified in the identification stage (Supplementary Table S6a), of which 126 lipids (91%) were quantified in both the discovery and identification stages

UHPSFC/MS measurements are repeated several months later to verify the repeatability of the results.

Part of the group separation between the cancer (red) group and the control (blue) group has been observed in the unsupervised PCA scoring chart, which confirms that the lipid group loss has a higher reproducibility, as shown in Figure 4A and B The number of selected samples is shown
.
Figures 4C-E further illustrate the reproducibility of the quantification results of the disordered lipids selected during the discovery and identification stage
.
In addition, the box plot also showed that compared with the control group, the selected lipids were down-regulated in all cancer types (Figure 4C-E)
.
The prediction results of these two stages are comparable
.

The ROC curves of all cancer types are shown in Figure 5A-D
.
The range of AUC value is 0.
888~0.
994
.
The specificity, sensitivity, and accuracy of a single model are as follows: 62%, 91%, and 76% of men with renal cancer: 86%, 72%, 83% of men with renal cancer: 94%, 57%, and 76% of women with breast cancer: prostate cancer 88%, 75% and 82%
.
Figure 5E-H is a summary of ROC curves of all samples of all types of cancer, with AUC values ​​ranging from 0.
864 to 0.
901
.

The ROC curves of all cancer types are shown in Figure 5A-D
.
The range of AUC value is 0.
888~0.
994
.
The specificity, sensitivity, and accuracy of a single model are as follows: 62%, 91%, and 76% of men with renal cancer: 86%, 72%, 83% of men with renal cancer: 94%, 57%, and 76% of women with breast cancer: prostate cancer 88%, 75% and 82%
.
Figure 5E-H is a summary of ROC curves of all samples of all types of cancer, with AUC values ​​ranging from 0.
864 to 0.
901
.

Figure 5I-L summarizes the ROC curves of the training set and validation set for all types of cancer
.
The value range of AUC is 0.
841 ~ 1.
00
.
For the two most significant lipids, the box plots of all pathological states are shown in Figure 5M, N, and the two-stage data sets of UHPSFC/MS (1,2) and shotgun MS (SG) are shown in Figure 5M, N.
Show
.

Figure 5I-L summarizes the ROC curves of the training set and validation set for all types of cancer
.
The value range of AUC is 0.
841 ~ 1.
00
.
For the two most significant lipids, the box plots of all pathological states are shown in Figure 5M, N, and the two-stage data sets of UHPSFC/MS (1,2) and shotgun MS (SG) are shown in Figure 5M, N.
Show

Taking into account all cancer types, methods and measurements, at least 10 of the 15 data subsets were corrected by Bonferroni, and 29 lipids were statistically significant
.
The accuracy decreases slightly with the different tumor types and methods, and the decrease in the amount of lipids used for MDA (Supplementary Tables S4 and S11)
.
However, since the decrease in prediction performance is not obvious, we studied the effect of further reduction of lipids in MDA on the prediction performance of correctly assigned sample types
.
According to Bonferroni correction, at least 14 data subsets of 15 data subsets considering all cancer types, methods and measurements are significant
.
MDA was performed for these 7 lipid species, and the prediction performance was evaluated
.
Taking into account all methods and genders, the average values ​​of sensitivity, specificity and accuracy of different amounts of lipids were calculated (Figure 6 )
.

Taking into account all cancer types, methods and measurements, at least 10 of the 15 data subsets were corrected by Bonferroni, and 29 lipids were statistically significant
.
The accuracy decreases slightly with the different tumor types and methods, and the decrease in the amount of lipids used for MDA (Supplementary Tables S4 and S11)
.
However, since the decrease in prediction performance is not obvious, we studied the effect of further reduction of lipids in MDA on the prediction performance of correctly assigned sample types
.
According to Bonferroni correction, at least 14 data subsets of 15 data subsets considering all cancer types, methods and measurements are significant
.
MDA was performed for these 7 lipid species, and the prediction performance was evaluated
.
Taking into account all methods and genders, the average values ​​of sensitivity, specificity and accuracy of different amounts of lipids were calculated (Figure 6 )
.
According to Bonferroni correction, at least 14 data subsets of 15 data subsets considering all cancer types, methods and measurements are significant
.
MDA was performed for these 7 lipid species, and the prediction performance was evaluated
.
Taking into account all methods and genders, the average values ​​of sensitivity, specificity and accuracy of different amounts of lipids were calculated (Figure 6 )
.

Generally speaking, the sensitivity, and therefore the accuracy, decreases, and the amount of lipids used for MDA decreases
.
This specificity is not affected by the number of lipid species and has nothing to do with the type of cancer
.
In all types of cancer, the effect of cancer stage on the concentration of the most important lipids was not observed (Supplementary Figure S2)
.

Generally speaking, the sensitivity, and therefore the accuracy, decreases, and the amount of lipids used for MDA decreases
.
This specificity is not affected by the number of lipid species and has nothing to do with the type of cancer
.
Generally speaking, the sensitivity, and therefore the accuracy, decreases, and the amount of lipids used for MDA decreases
.
This specificity is not affected by the number of lipid species and has nothing to do with the type of cancer
.
In all types of cancer, the effect of cancer stage on the concentration of the most important lipids was not observed (Supplementary Figure S2)
.

Statistical evaluation of data
.

Statistical evaluation of data
.

By evaluating the statistically significant lipid species after Bonferroni correction, the lipid species with a fold change of ±20% and VIP value >1 , the characteristics of different plasma lipids were studied
.
According to the pie charts of different cancer types in Figs.
7A-D, the percentage of lipid species belonging to lipids that meet the defined criteria is calculated
.
Non-polar lipids, triacylglycerols and cholesterol esters, have greater relevance in kidney cancer, while the effects of glycerophospholipids and sphingolipids seem to be dominant in breast and prostate cancer
.

By evaluating the statistically significant lipid species after Bonferroni correction, the lipid species with a fold change of ±20% and VIP value >1 , the characteristics of different plasma lipids were studied
.
Lipid species> 1 According to the pie charts of different cancer types in Figure 7A-D, the percentage of lipid species belonging to lipids that meet the defined criteria is calculated
.
Non-polar lipids, triacylglycerols and cholesterol esters, have greater relevance in kidney cancer, while the effects of glycerophospholipids and sphingolipids seem to be dominant in breast and prostate cancer
.
Non-polar lipids, triacylglycerols and cholesterol esters, have greater relevance in kidney cancer, while the effects of glycerophospholipids and sphingolipids seem to be dominant in breast and prostate cancer
.
cholesterol

As shown in Figure 7E-G, in the plasma samples of cancer patients, the most important lipids in all methods and data sets were down-regulated, regardless of the type of cancer
.
Through MDA to study the difference of all quantitative lipids in healthy control samples and different types of cancer samples, the difference of 91 lipids in UHPSFC/MS and shotgun MS, the difference of 29 lipids, and finally, the 7 most significant differences in all data sets Lipids
.
Healthy male control group and OPLS-DA model of prostate cancer, kidney cancer patient samples and healthy male and female control group kidney cancer, samples of women with breast cancer FIG.
8A, B shown in FIG
.
The question is whether it is possible to differentiate and predict cancer types and healthy control samples using UHPSFC/MS
.

As shown in Figure 7E-G, in the plasma samples of cancer patients, the most important lipids in all methods and data sets were down-regulated, regardless of the type of cancer
.
Through MDA to study the difference of all quantitative lipids in healthy control samples and different types of cancer samples, the difference of 91 lipids in UHPSFC/MS and shotgun MS, the difference of 29 lipids, and finally, the 7 most significant differences in all data sets Lipids
.
Through MDA to study the difference of all quantitative lipids in healthy control samples and different types of cancer samples, the difference of 91 lipids in UHPSFC/MS and shotgun MS, the difference of 29 lipids, and finally, the 7 most significant differences in all data sets Lipids
.
Healthy male control group and OPLS-DA model of prostate cancer, kidney cancer patient samples and healthy male and female control group kidney cancer, samples of women with breast cancer FIG.
8A, B shown in FIG
.
The question is whether it is possible to differentiate and predict cancer types and healthy control samples using UHPSFC/MS
.
Is it possible to use UHPSFC/MS to differentiate and predict cancer types and healthy control samples
.

For training and validation sets and different amounts of lipids (138, 91, 29, and 7 lipids), the specificity ranges from 97% to 67 with an average of 83%, and the renal cancer sensitivity is an average of 49-74% 61%; the sensitivity of prostate cancer ranges from 0 to 66%, with an average of 43%, and the accuracy ranges from 57 to 74%, with
an average of 65%, to construct an MDA model of male samples (Figure 8A) .
Specificity ranges from 96% to 80, with an average of 90%, renal cancer sensitivity from 0 to 44% with a mean of 24%, breast cancer sensitivity from 60 to 83% with a mean of 73%, and accuracy from 78% to 66 with a mean Construct a female sample MDA model
for 74% of the training (Figure 8B) .
Due to the insufficient number of sample components obtained from female donors, it is impossible to use 7 lipids as variables for the MDA model
.
The OPLS-DA model classifies kidney cancer and prostate cancer in men (Figure 8C), and classifies kidney cancer and breast cancer in women (Figure 8D)
.

For training and validation sets and different amounts of lipids (138, 91, 29, and 7 lipids), the specificity ranges from 97% to 67 with an average of 83%, and the renal cancer sensitivity is an average of 49-74% 61%; the sensitivity of prostate cancer ranges from 0 to 66%, with an average of 43%, and the accuracy ranges from 57 to 74%, with
an average of 65%, to construct an MDA model of male samples (Figure 8A) .
Specificity ranges from 96% to 80, with an average of 90%, renal cancer sensitivity from 0 to 44% with a mean of 24%, breast cancer sensitivity from 60 to 83% with a mean of 73%, and accuracy from 78% to 66 with a mean Construct a female sample MDA model
for 74% of the training (Figure 8B) .
Due to the insufficient number of sample components obtained from female donors, it is impossible to use 7 lipids as variables for the MDA model
.
The OPLS-DA model classifies kidney cancer and prostate cancer in men (Figure 8C), and classifies kidney cancer and breast cancer in women (Figure 8D)
.

For the training set and validation set and the UHPSFC/MS data set of male samples, the sensitivity to prostate cancer is 71-88%, with an average of 81%, and the sensitivity to kidney cancer is 57-82%, with an average of 72%
.
For the training set, validation set and UHPSFC/MS data set of female samples, the sensitivity to breast cancer is between 100% and 94, with an average of 98%, and the sensitivity is 57% with a mean ranging from 14 to 86% for kidney cancer.
, And the accuracy is between 97% and 80, with an average of 88%
.

For the training set and validation set and the UHPSFC/MS data set of male samples, the sensitivity to prostate cancer is 71-88%, with an average of 81%, and the sensitivity to kidney cancer is 57-82%, with an average of 72%
.
For the training and validation sets and female samples UHPSFC / MS data set, pair of pair of sensitivity between 100 Breast 94% and an average of 98%, a sensitivity of renal 14--86% range of the average 57 %, and the accuracy is between 97% and 80, with an average of 88%
.
The sensitivity of breast cancer is between 100% and 94, with an average of 98%, the sensitivity is 57% with an average of 14-86% for kidney cancer, and the accuracy is between 97% and 80, with an average of 88%
.

In general, this study studied the lipidome analysis of human plasma through the combination of ultra-high performance supercritical fluid chromatography and mass spectrometry, with the purpose of distinguishing breast, kidney, and prostate cancer patients from healthy controls
.
The average sensitivity, specificity, and accuracy of the lipid mapping method for renal cancer are 85%, 95%, and 92%, respectively ; breast cancer are 91%, 97%, and 94%, respectively ; prostate cancer are 87%, 95%, and 95%, respectively.
92%, there is no correlation between statistical model and tumor stage
.
Statistically the most significant lipids for cancer differentiation are CE 16:0, Cer 42:1, LPC 18:2, PC 36:2, PC 36:3, SM 32:1, and SM 41:1, these 7 This lipid is a potential biomarker for kidney, breast, and prostate cancer screening, but it needs further verification in prospective studies to verify its clinical utility
.

In general, this study studied the lipidome analysis of human plasma through the combination of ultra-high performance supercritical fluid chromatography and mass spectrometry, with the purpose of distinguishing breast, kidney, and prostate cancer patients from healthy controls
.
The average sensitivity, specificity, and accuracy of the lipid mapping method for renal cancer are 85%, 95%, and 92%, respectively ; breast cancer are 91%, 97%, and 94%, respectively ; prostate cancer are 87%, 95%, and 95%, respectively.
92%, there is no correlation between statistical model and tumor stage
.
Statistically the most significant lipids for cancer differentiation are CE 16:0, Cer 42:1, LPC 18:2, PC 36:2, PC 36:3, SM 32:1, and SM 41:1, these 7 This lipid is a potential biomarker for kidney, breast, and prostate cancer screening, but it needs further verification in prospective studies to verify its clinical utility
.

 

Original source:

Original source:

DeniseWolrab, et al.
Plasma lipidomic profles of kidney,breast and prostate cancer patients difer from healthy controls.

DeniseWolrab, et al.
Plasma lipidomic profles of kidney,breast and prostate cancer patients difer from healthy controls.

Scientifc Reports | (2021) 11:20322 | https://doi.
org/10.
1038/s41598-021-99586-1

Leave a message here