p53 immunostaining pattern is a useful surrogate marker for TP53 gene mutations

Abstract

Background

TP53 is the most frequently mutated gene in the human cancer, and the awareness of its mutational status is useful in the diagnosis and treatment of cancer patients. In the present study, we investigated the association between TP53 gene mutations and p53 immunohistochemical staining (IHC) patterns and non-genetic effect of MDM2 as a negative regulator of p53.

Methods

A total of 135 solid cancer cases with next generation sequencing data were subjected to p53 IHC and classified as overexpression, null type or usual pattern.

Results

TP53 mutation was observed in 104 out of 135 cases (77.0%). When the TP53 mutations were annotated into DISRUPTED (truncations, frameshifts, splice site mutations, and deep deletions) and IF-DBD (in-frame mutations in the DNA binding domain), the null type p53 IHC pattern was associated with DISRUPTED mutations (sensitivity 86.2%, specificity 97.2%) while the overexpression pattern was associated with IF-DBD mutations (sensitivity 100%, specificity 81.7%). The specificity of p53 IHC usual pattern predicting wild type TP53 was also as high as 100%. Regardless of MDM2 amplification, p53 IHC pattern showed a perfect association with TP53 mutation pattern.

Conclusions

p53 IHC pattern (overexpression, null type, usual) reasonably predicted TP53 mutational status (DISRUPTED, IF-DBD), and MDM2 amplification status did not have any impact on the p53 IHC pattern.

Background

TP53, functioning as a cellular stress sensor for physiologic or oncogenic activation, is the most frequently mutated gene in human cancer [1,2,3,4]. When there is oncogenic activation, DNA replication stress disrupts physiologic function of p53, resulting in cell death evasion and thus, TP53 mutation is selected during tumor progression [1, 5,6,7]. Because the p53 is so essential for tumorigenesis and tumor progression, there have been several attempts to elicit tumor cell death by reactivation of mutated TP53 [8,9,10,11]. For instance, a conceptual basis for alleviating the effect of inactivating TP53 mutation through artificial second-site mutation has been proposed [12, 13]. Furthermore, pathogenic TP53 mutations are known to be associated with poor survival or chemo-resistance in various tumor types [12, 14,15,16,17,18,19,20,21]. Therefore, it is very important to check the TP53 mutational status to predict the patient’s prognosis and treatment responsiveness.

Sanger sequencing is known as a traditional gold standard method to confirm the mutational status of TP53, but it’s labor intensive and time consuming especially because the TP53 gene is quite big. Because of these practical problems, it’s difficult to use Sanger sequencing in actual practice, and instead, immunohistochemical staining (IHC) is the most widely used way to infer TP53 mutational status. Several attempts have been made to infer TP53 mutation type through p53 IHC pattern, but most previous studies were limited to one specific organ, and the methods of interpreting the p53 IHC results varied, such as pattern-based or positive cell proportions [22,23,24,25,26,27,28,29,30].

To better understand the association between TP53 gene mutations and p53 IHC patterns, we used a set of tumor samples having TP53 gene mutation data across entire TP53 exons obtained by a clinical grade next-generation sequencing (NGS) test. Through annotation of TP53 mutations according to the predicted effects on p53 protein function and classification of p53 IHC patterns according to the presence of overexpression or complete loss of expression, we investigated the association between the TP53 mutation status and p53 IHC pattern. When the p53 IHC pattern and expected TP53 mutation status were discordant, we analyzed the reasons for the discrepancies. Secondly, we investigated p53 IHC pattern in MDM2-amplified cases to determine how p53 IHC appears through the non-genetic effect of MDM2, a negative regulator of p53.

Methods

Case selection

After the approval (protocol number, 2018–0976) from the institutional review board, we retrospectively collected 135 adult solid cancer samples that have clinical next-generation sequencing (NGS) data from the records of the Department of Pathology at the Asan Medical Center, University of Ulsan College of Medicine. Among them, 26 cases were obtained after neoadjuvant treatment and 12 cases showed unequivocal MDM2 amplification.

Mutational analysis

The clinical NGS cancer panel test was performed as described previously [31]. Briefly, the targeted NGS used the MiSeq platform (Illumina, San Diego, CA) and the gene panel, OncoPanel AMC version 3, was designed in house through SureDesign (Agilent Technologies, Santa Clara, CA) using GRCh37 reference version. This 1.2 Mbp-sized panel included 33,524 probes targeting a total of 382 genes, including entire exons of 199 genes, 184 hot spots, and partial introns for eight genes often rearranged in cancer. Of course, the entire TP53 and MDM2 exonic regions were included.

We classified the TP53 mutations according to the type of nucleotide changes: single nucleotide variation (SNV), in-frame insertion or deletion (In-frame indel), frameshift insertion of deletion (frameshift indel), premature stop codon (truncation), splice site mutation (splice), and copy-number (CN) loss. We classified the TP53 mutations into two major categories, in-frame alterations across the DNA binding domain (IF-DBD) and alterations involving significant disruption of protein coding sequences (DISRUPTED). The IF-DBD mutations included SNVs and In-frame indel across DNA binding domain. The DISRUPTED mutations included frameshift indel, truncation, splice, and CN loss. The MDM2 amplifications were detected through our bioinformatics pipeline using CNVkit v0.9.6 [32]. Copy numbers of tumors were called against a panel of normal. Copy number segments were inferred using circular binary segmentation(CBS) method. Absolute copy number of given segments was estimated using ‘call’ module of CNVkit and pathologist’s tumor cellularity. We used pathologist’s tumor purity estimates to infer copy-number more accurately. This involved solving an equation: (measured copy number) = (actual copy number) * (tumor purity) + 2 * (1-tumor purity). To investigate the biologic effect of MDM2 overexpression on TP53 status, we included tumor samples with high level MDM2 amplifications (estimated MDM2 copy number > 30). All cases were manually reviewed through visualization of bam files with the integrated genomic browser (IGV) and double check of tumor cell purity through review of microscopic slides and variant allelic fractions of the detected clonal mutations.

Immunohistochemistry

Representative areas from formalin-fixed, paraffin-embedded tissues were subjected to IHC, which was performed on 4-μm-tick sections using a Ventana auto-stainer and an ultra-View DAB Detection Kit (Ventana, Tucson, Arizona), according to the manufacturer’s instructions. Primary antibody for p53 (clone DO-7, catalog No.M7001, DAKO, Denmark, Glostrup, 1:1000, Mouse monoclonal) was used. The p53 IHC staining were evaluated by two pathologists (JK and YNS) and classified into 3 categories: overexpression (strong diffuse nuclear immunoreactivity in all tumor cells), null (complete absence of nuclear immunoreactivity in all tumor cells), and usual (neither overexpression nor null, variable nuclear immunoreactivity in tumor cells).

Statistical analysis

The R software (version 4.02 Vienna, Austria) was used to perform statistical analyses. Association between TP53 mutational status by NGS test and IHC staining was tested using the χ2 and/or the Fisher exact tests. P-values less than 0.05 were considered statistically significant.

Results

Malignant tumors of various organs were included in this study: anus in 2 (1.5%), brain in 7 (5.2%), breast in 6 (4.4%), large intestine in 107 (79.3%), lung in 2 (1.5%), ovary in 6 (4.4%), small intestine in 1 (0.7%), stomach in 2 (1.5%), and unknown in 2 (1.5%) cases (Table 1). Pathologic diagnosis of tumor was mostly adenocarcinoma (113 cases, 83.7%). Most cases were surgically resected specimens (125 cases, 92.6%) followed by forceps or needle biopsies (5 cases, 3.7%), and excisional biopsies (5 cases, 3.7%). Before obtaining tissue, neoadjuvant treatment was performed in 26 cases (19.3%).

Table 1 Clinicopathologic characteristics of cases

Through NGS analysis, TP53 mutation was observed in 104 out of 135 cases (77.0%). Of the 104 cases with the TP53 mutation, 72 types of TP53 mutations were observed (Supplementary data). Table 2 shows the association of TP53 mutational type and the IHC staining pattern. As a result of p53 IHC pattern, there were 86, 28 and 21 cases of overexpression, null, and usual pattern, respectively (Fig. 1). Most of the cases with p53 overexpression pattern except for 7 cases were accompanied by mutations (79/86, 91.9%) such as single nucleotide variation, In-frame indel, truncation, CN loss, and splice. Of the 7 cases without mutation, 4 cases received neo-adjuvant chemotherapy. Similarly, in the cases of p53 with null pattern, 25 out of 28 cases (89.3%) accompanied mutations. No mutations were observed in all cases (21/21, 100%) showing the p53 usual pattern.

Table 2 Immunohistochemical staining patterns and mutational analysis of TP53

Fig. 1

Representative figures for three p53 immunohistochemistry patterns. A Invasive ductal carcinoma of the breast harboring TP53 H193R mutation (Hematoxylin & Eosin, × 1.25 objective lens). B Diffuse strong immunoreactivity for p53, that is overexpression pattern (p53 immunohistochemistry, × 1.25 objective lens). C Low grade papillary serous carcinoma of the ovary without any oncogenic TP53 mutation (Hematoxylin & Eosin, × 10 objective lens) D Approximately a half of the tumor cells express p53 in the nuclei, that is usual pattern (p53 immunohistochemistry, × 10 objective lens). E Adenocarcinoma of the colon harboring TP53 H214Qfs*2 mutation (Hematoxylin & Eosin, × 4 objective lens). F Tumor cells are completely negative for p53 protein expression. Adjacent non-neoplastic cells showing p53 expression serves as an internal positive control. This pattern is classified as null pattern (p53 immunohistochemistry, × 4 objective lens)

The association of TP53 mutation type and p53 IHC pattern was better observed when the TP53 mutation were divided into IF-DBD and DIRUPTED according to the resulting sequence context (Table 3, p < 0.001). Because the in-frame SNVs or Indels are predicted to maintain most of the protein coding sequences intact while other alterations may disrupt amino acid sequences considerably, the p53 protein IHC pattern is predicted to be different between the two groups. Indeed, overexpression pattern was closely associated with the IF-DBD TP53 mutations (sensitivity 100%, specificity 81.7%, positive predictive value 87.2%, negative predictive value 100%), while null pattern was associated with the DIRUPTED TP53 mutations (sensitivity, 86.2%, specificity 97.2%, positive predictive value 89.3%, negative predictive value 96.3%). The usual p53 IHC pattern well predicted the absence of pathogenic TP53 mutations (sensitivity 67.7%, specificity 100%, positive predictive value 100%, negative predictive value 91.2%). Additionally, among 26 cases that received previous neoadjuvant treatment, there were 19, 6, and 1 cases of overexpression, null type, and usual pattern, respectively. And the association between TP53 mutation type and p53 IHC pattern was statistically significant (Table 4, p < 0.002).

Table 3 Immunohistochemical staining patterns and predicted functional consequence of TP53 mutations

Table 4 Immunohistochemical staining patterns and predicted functional consequence of TP53 mutations within cases with previous neoadjuvant treatment

Table 5 shows the p53 IHC pattern according to the MDM2 amplification status. The MDM2-amplified tumors showed more frequent usual p53 IHC pattern than MDM2-non-amplified tumors (p < 0.001, Fisher’s exact test). Null pattern was observed in only one case (1/12, 8.3%) in contrast to the assumption that p53 may be downregulated through MDM2 amplification. Instead, when we looked at the presence and the type of TP53 mutations in the MDM2-amplified tumors, the p53 IHC pattern just followed TP53 mutation status regardless of MDM2 amplification except for one case where p53 IHC overexpression pattern and wild type TP53 mutation was observed (Table 6).

Table 5 Immunohisochemical staining pattern of p53 according to MDM2 amplification status

Table 6 Immunohistochemical staining patterns according to the TP53 mutation types in MDM2-amplified cases

The p53 IHC pattern and TP53 mutation discordance was observed in 14 out of 135 cases (14/135, 10.4%) (Table 7), and could be divided into the following 2 groups; 1) Wild type TP53 gene showing either overexpression or null IHC pattern, and 2) DISRUPTED type TP53 mutation showing overexpression pattern. In group 1, 7 cases showed overexpression and 3 cases showed null p53 IHC pattern in the absence of TP53 mutations. Among the 7 cases showing p53 overexpression, 4 cases received neo-adjuvant chemoradiation therapy (Fig. 2). In addition, 2 out of 3 cases with p53 IHC null also received neo-adjuvant chemoradiation therapy. Since group 1 type discrepancies could also be seen due to failure of TP53 mutation detection, we re-checked tumor cell purity. However, the tumor cell purity was within acceptable range in all cases (Fig. 3). The other cases belonged to the group 2: DISRUPTED type TP53 mutation and overexpression p53 IHC pattern.

Table 7 p53 immunohistochemical staining pattern and TP53 mutation discordant cases

Fig. 2

Mucinous adenocarcinoma of the colon that experienced neo-adjuvant chemoradiation therapy but did not show any TP53 mutation. A Several strips or clusters of tumor cells are floating in the mucin pool (Hematoxylin & Eosin, × 10 objective lens). B TP53 overexpression pattern is noted (p53 immunohistochemistry, × 10 objective lens)

Fig. 3

Representative case with discrepancy between TP53 mutation and p53 IHC pattern. A&B Tumor area from which DNA was extracted for NGS analysis. Tumor cell purity is in acceptable range (Hematoxylin & Eosin, A: X4, B: X10 objective lens, respectively). C Null type pattern of p53 immunostaining is observed. No TP53 mutation was detected in this case (p53 immunohistochemistry, × 10 objective lens). D, E Adenocarcinoma of the cecum with comedo-type necrosis. Area from which DNA has been extracted is shown in E (Hematoxylin & Eosin, D: X1.25, and E: X4 objective lens, respectively). F Overexpression pattern of p53 immunostaining is noted in the absence of detectable TP53 mutation (p53 immunohistochemistry, × 4 objective lens). G, H Moderately differentiated adenocarcinoma of the rectum. DNA has been extracted from the area shown in G (Hematoxylin & Eosin, G: X4, and H: X10 objective lens, respectively). I Overexpression pattern of p53 immunostaining is noted in the absence of detectable TP53 mutation. This patient did not receive neoadjuvant chemoradiation therapy (p53 immunohistochemistry, × 10 objective lens)

Discussion

In this study, the presence of oncogenic or likely oncogenic TP53 mutations and their predicted functional consequences were well correlated with p53 IHC pattern in most cases. Sensitivity for p53 overexpression or null type pattern for prediction of IF-DBD type or DISRUPTED type TP53 mutations was 100% and 86.2%, respectively, and specificity for TP53 usual pattern for prediction of WT TP53 status was also high as 100%. MDM2 amplification did not significantly affect TP53 IHC pattern, in contrast to the expectation that MDM2 would act as a negative regulator of p53, resulting in loss of p53 protein expression. Taken together, the p53 IHC pattern may serve as a reasonable surrogate for TP53 mutations.

For the exceptions in those associations, some cases could be explained by predicted biological consequences. The 4 out of 7 cases with p53 overexpression in the absence of any TP53 mutation received neo-adjuvant chemoradiation therapy. As physiologic p53 accumulation is possible due to the neo-adjuvant chemoradiation therapy, p53 overexpression may be observed in the absence of TP53 gene mutations. However, the other occasions, such as p53 IHC null pattern in tumors harboring wild type TP53, p53 IHC overexpression pattern in tumors harboring DISRUPTED type TP53 mutations, or p53 overexpression in chemoradiation-naïve tumors with wild type TP53, are hard to explain. There might be additional factors affecting p53 IHC pattern other than the presence of the type of TP53 mutations that require further investigation.

As for the association between types of TP53 mutations and p53 IHC patterns, IF-DBD TP53 mutations were commonly associated with overexpression pattern while DISRUPTED TP53 mutations were associated with null pattern. Those associations have been explained by the accumulated biological experiment data that p53 protein encoded by IF-DBD TP53 mutations accumulate in the tumor cell nuclei through interfered MDM2-mediated ubiquitination, and DISRUPTED TP53 mutations result in non-sense mediated decay, or prematurely truncated mRNA, and thus null pattern of p53 IHC [22]. Of note, MDM2 amplification was not associated with p53 IHC null pattern, suggesting that TP53 mutation itself may have a greater influence on IHC pattern than the role of MDM2 as a p53 negative regulator.

There have been several studies that attempted to correlate p53 IHC pattern with TP53 mutational type [22, 23, 29, 30]. In the study of Yemelyanova et al. [23], the results of p53 IHC were divided into 5 groups according to the percentage of positive tumor cells, and when positive tumor cells were more than 60% or 0%, the sensitivity to detect TP53 mutation was 94%. Additionally, in the study of Kobel et al. [22], the predictability of four types of p53 antibodies were evaluated by applying the same three-tier scoring system as in our study. Although it differed slightly depending on the type of antibody, the sensitivity and specificity for p53 IHC were 0.87 ~ 0.95 and 0.73 ~ 0.95 respectively. Singh et al. [29] also included cytoplasmic p53 IHC result as a mutant pattern, and the sensitivity and specificity for TP53 mutation were 97.7% and 88.89% respectively. These studies, together with our study, showed that p53 IHC is useful for the prediction of TP53 mutation type although not perfect.

Previous studies were mainly limited to one organ, but in this study, we could investigate p53 IHC patterns and TP53 mutations in various organs, including colon, brain, breast, ovary, lung, stomach and small intestine, by virtue of NGS. In addition, we also reviewed discrepant cases and the reasons for the discrepancies. Although some discrepancies could not be reasonably explained, we found that neoadjuvant chemoradiation may induce p53 overexpression in the absence of IF-DBD type TP53 mutations. Our study has some limitations. First, there may be selection bias due to availability of NGS data. In addition, the p53 IHC pattern was evaluated by only one type of monoclonal antibody, and detailed mechanistic study for discrepancies between p53 IHC pattern and TP53 mutation type could not be performed because of shortage of time and resources.

In conclusion, TP53 mutational status can be well predicted through three p53 IHC patterns (overexpression, null, and usual) in most cases: 1) overexpression for IF-DBD type TP53 mutations, 2) null for DISRUPTED type TP53 mutations, and 3) usual for the absence of TP53 mutations. In addition, MDM2 amplification does not seem to significantly affect the p53 IHC pattern. Finally, we propose that pattern-based approach (overexpression, null, and usual patterns) might be more informative in the interpretation of p53 IHC than traditional interpretation system based on positive tumor cell proportion.