Main

Among the subtypes of acute myeloid leukemia, acute myeloid leukemia with multilineage dysplasia represents a particularly aggressive subtype with a high mortality and poor response to treatment.1, 2, 3 Although acute myeloid leukemia with myelodysplasia-related changes is defined by particular clinical, morphologic, and cytogenetic features, there remains significant heterogeneity in clinical behavior and response to therapy among patients with this disease. For this reason, continued efforts to identify prognostic and predictive features are necessary to guide clinical decisions, including decisions regarding intensity of therapy.

The tumor suppressor TP53 is the most commonly mutated gene in human cancers4 and mutations in TP53 have been demonstrated in several studies to correlate with inferior overall survival among patients with acute myeloid leukemia and myelodysplastic syndrome.5, 6, 7, 8 In addition, TP53 mutations are found in ~70% of cases with a complex karyotype, while in cases with normal karyotype, mutations are comparatively rare.9 Although genotyping TP53 in clinical specimens has become relatively common, immunohistochemical staining for p53 has been shown to correlate with the presence of missense mutations in TP53 in several tumor types, and thus has some utility in identifying malignancies with TP53 mutations which have a worse prognosis.10, 11 Expression of p53 in acute myeloid leukemia has been examined previously,12, 13 but limited data are available regarding p53 expression in different World Health Organization acute myeloid leukemia subtypes. In therapy-related myeloid neoplasms, previous work has shown that overexpression of p53 correlates with TP53 mutation and independently predicts a decreased overall survival.14 In acute myeloid leukemia arising from myelodysplastic syndrome, p53 was found to be expressed in ≥5% of cells in 9 of 39 cases.15 In this same study, a p53 expression level of ≥5% did not stratify higher-risk myelodysplastic syndrome and secondary acute myeloid leukemia into groups with different overall survival. Further work is necessary to evaluate the utility of p53 immunohistochemistry in other acute myeloid leukemia subtypes.

In this study, we set out to correlate expression of p53 with acute myeloid leukemia subtype, karyotype, TP53 genotype, and survival in a cohort of 143 cases of acute myeloid leukemia. We find that cases of acute myeloid leukemia with a high level of p53 expression are typically acute myeloid leukemia- myelodysplasia-related changes as opposed to acute myeloid leukemia–not otherwise specified or acute myeloid leukemia with recurrent genetic abnormalities. In addition, cases with higher p53 expression frequently have TP53 mutations, complex karyotypes and show a worse overall survival. Our data suggest that p53 immunohistochemistry may provide useful prognostic information in cases of acute myeloid leukemia with myelodysplasia-related changes.

Materials and methods

Case Selection and Immunohistochemistry

We identified 143 cases of acute myeloid leukemia from the Stanford Department of Pathology (obtained from 12 May 2004 to 25 October 2013) for which the original diagnostic bone marrow core biopsy was available. Cases were classified according to the most recently available 2016 World Health Organization categories.1, 3 After fixation in Bouin solution and decalcification in Formical-4 (StatLab, McKinney, TX, USA), sections of the paraffin-embedded core biopsy were stained with an antibody to p53 (clone DO-7 mouse, Ventana, 1:400 dilution, Ventana CC1 retrieval solution, pH 6.0). For the entire cohort, core biopsies were considered adequate for evaluation if there were at least ~1000 cells present for evaluation and this was the case even for biopsies with hypocellular marrow. This study was approved by Stanford University’s institutional review board.

Method for Quantifying p53 Expression

Examination of the 143 cases revealed that the majority of the core biopsies showed only rare (<1%) nuclei with dim to moderate expression of p53. These cases also tended to be those that lacked a detectable TP53 mutation (discussed in detail in the Results section). In contrast, a subset of cases, which tended to have nonsynomous TP53 mutations, showed a higher proportion of p53-positive nuclei with intensity ranging from dim to bright. Given that the background level of expression of p53 in TP53 WT cases was very low in terms of the proportion of positive cells as well as intensity, we reasoned that an increased proportion of even dimly positive cells was significant. For this reason, we chose to count dimly positive cells in all of the counts used in this study. We also think that counting every nucleus that is at least dimly positive may have the added benefit of reducing the interobserver variability compared with choosing a threshold at a moderate level of intensity. The p53-stained core biopsies were evaluated and the percent of all nucleated cells with at least 1+ positive nuclei (defined as above background as compared with negative cases) was determined by manual counting. In core biopsies with variability in p53 staining in different fields, hotspots were counted. After close evaluation, it was decided that precise quantification of p53 staining in cases in the 0–5% (n=119) group could not be reliably subcategorized by manual counting without a high degree of interobserver variability. The cases in which p53 expression was greater than 5% were precisely quantified by taking pictures of two optical fields at 200x magnification, printing images of the fields and manually counting the number of p53-positive nuclei. The average total number of cells counted per case was 1790 and the median was 1860 (range 1000–2925). For 15 cases with p53 score >5%, a second observer counted at least 1000 cells to determine the percent of cells expressing at least dim nuclear p53. A linear regression comparing the two observers resulted in a line with an R2=0.97 (Supplementary Figure 1).

TP53 Sequencing

For 67 of the cases, unstained aspirate smears or frozen bone marrow samples were available to evaluate the TP53 gene by Sanger sequencing to detect the presence of known frequently occurring mutations. Targeted sequencing of the TP53 gene was performed as described previously16 with three primer sets: Set 1 Forward 5′-TCTGTCTCCTTCCTCTTCCTAC-3′ and Reverse 5′-CTGCTCACCATCGCTATCTG-3′; Set 2 Forward 5′-CACATGACGGAGGTTGTGAG-3′ and Reverse 5′-TAGGGCACCACCACACTAT-3′; Set 3 Forward 5′-CCTGATTTCCTTACTGCCTCTT-3′ and Reverse 5′-TCTTGTCCTGCTTGCTTACC-3′.

Statistical Analysis

Statistical tests used in this study are indicated in the text and in the figure legends and were performed using SPSS software (SPSS, Chicago, IL, USA) and/or R software (https://cran.r-project.org/). Cox proportional hazard model was used for univariate and multivariate analyses for overall survival.

Results

Patient Cohort

A summary of cohort characteristics is provided in Table 1. The average age of the study cohort was 55.9 years old (median 60, range 5–80) and consisted of 50% females. Of the 143 cases, 71 (50%) fulfilled the WHO 2008 criteria for acute myeloid leukemia with myelodysplasia-related changes, 40 (28%) were acute myeloid leukemia–not otherwise specified, 25 (18%) were acute myeloid leukemia with recurrent genetic abnormalities, and 7 (5%) were therapy-related acute myeloid leukemia.

Table 1 Summary of cases evaluated in this study
Full size table

p53 Expression in 143 Cases of AML

Of the 143 cases, 24 (17%) contained more than 5% of the cells in the marrow with nuclear p53 expression above background. Among the 119 cases that we categorized as having ≤5% p53-positive cells, cases either showed no definite p53-positive cells or at most fewer than 5% nuclei with very dim staining. Within each case with >5% p53-positive nuclei, the staining intensity ranged from very dim to very bright. Examples of p53-stained core biopsies, together with relevant clinical information, are provided in Figure 1.

Figure 1
figure 1

Representative core biopsies stained with p53 antibody. (a) Acute myeloid leukemia–not otherwise specified, TP53 not sequenced, 12.5% p53-positive cells by manual count. Patient alive at 1771 days. (b) Acute myeloid leukemia–not otherwise specified with normal karyotype, TP53 not sequenced, patient alive at 319 days, 4% p53-positive cells by manual count. (c) Acute myeloid leukemia with myelodysplasia-related changes with complex karyotype, TP53 not sequenced, 16% positive cells, patient deceased at 10 days. (d) Acute myeloid leukemia with myelodysplasia-related changes with complex karyotype, TP53 not sequenced, 19% p53-positive cells, patient deceased at 102 days. (e) Acute myeloid leukemia with myelodysplasia-related changes with complex karyotype, TP53 not sequenced, 26% p53-positive cells, patient deceased at 247 days. (f) Acute myeloid leukemia with myelodysplasia-related changes with complex karyotype with TP53 mutation (C176F) detected, 79% p53-positive cells, patient alive at 286 days. (g) Acute myeloid leukemia with myelodysplasia-related changes with complex karyotype, frameshift TP53 mutation, clinical follow up not available. (h) Acute myeloid leukemia with myelodysplasia-related changes (biphenotypic) with complex karyotype, only rare positive cells, no TP53 mutation detected, patient alive at 234 days. (i) Acute myeloid leukemia with myelodysplasia-related changes with complex karyotype, R273H TP53 mutation detected, 14% p53-positive cells, clinical follow up not available. (j) Therapy-related acute myeloid leukemia with normal karyotype with rare strongly positive cells, no TP53 mutation detected, patient deceased at 189 days. (k) Acute myeloid leukemia with myelodysplasia-related changes normal karyotype with 11% p53-positive cells, no TP53 mutation detected, clinical follow up not available. (l) Acute myeloid leukemia with myelodysplasia-related changes complex karyotype with rare dim positive cells, R196* TP53, patient deceased at 594 days. (m) Acute myeloid leukemia–not otherwise specified with WT TP53, alive at 423 days of follow up with 46% p53-positive cells. (n) Acute myeloid leukemia–not otherwise specified with TP53 V274F mutation and 7% p53-positive cells. (o) Acute myeloid leukemia with recurrent genetic abnormalities with mutated NPM1, wild-type TP53, 15% p53-positive cells.

Full size image

On the basis of the morphologic features, it appeared that the majority of the positive cells were blasts. In 5 cases, p53 expression was seen in megakaryocyte nuclei (Supplementary Table 4). Four of these cases were acute myeloid leukemia with myelodysplasia-related changes and one was therapy-related acute myeloid leukemia. All of these cases had a complex karyotype and the proportion of p53-positive cells in the 5 cases ranged from 16 to 53%. TP53 sequence data was available for the therapy-related acute myeloid leukemia case, which showed a mutation leading to an R175H substitution.

On the basis of our evaluation of the stained biopsies, it appeared that distinction of a true p53 ‘null’ phenotype, characterized by a complete absence of detectable p53 expression, was difficult to discern from cases with a very low level of p53 expression. In addition, our cohort contained only two cases in which the TP53 genotype might predicted to yield a p53 null phenotype due to loss of a significant portion of the WT protein. Both of these cases were acute myeloid leukemia with myelodysplasia-related changes and both showed very few p53-positive cells and included 1 case with a frameshift mutation (shown in Figure 1g) and 1 case with a nonsense mutation (shown in Figure 1l). Our conclusion was that it was difficult or impossible to reliably distinguish these cases from the numerous other cases with less than 5% p53-positive cells that were TP53 WT.

Because of the difficulty in accurately quantifying p53 expression at low levels, we grouped cases with less than or equal to 5% p53-positive cells into one bin, which contained 119 of the cases. We grouped cases with >5% expression into bins with intervals of 6–10%, 11–15%, 15–20%, 21–25%, etc. Of the 24 cases with >5% p53-positive cells, 17 were acute myeloid leukemia with myelodysplasia-related changes (71%), 5 were acute myeloid leukemia–not otherwise specified (21%), 1 was therapy-related acute myeloid leukemia (4%), and 1 (4%) was acute myeloid leukemia–with recurrent genetic abnormalities. The distribution of p53 expression for all of the cases, as well as for each acute myeloid leukemia subcategory, is shown in Figure 2.

Figure 2
figure 2

Percent of cases in each category with the indicated level of p53-positive cells. AML-MRC, acute myeloid leukemia with myelodysplasia-related changes, AML-NOS, acute myeloid leukemia–not otherwise specified, AML-RGA, acute myeloid leukemia with recurrent genetic abnormalities, AML-T, therapy-related acute myeloid leukemia.

Full size image

Five Cases of Acute Myeloid Leukemia–Not Otherwise Specified and 1 Case of Acute Myeloid Leukemia With Recurrent Genetic Abnormalities With Elevated p53 Expression

While cases with an elevated proportion of p53-positive cells were more frequently seen in the acute myeloid leukemia with myelodysplasia-related changes subcategory, 5 cases of acute myeloid leukemia–not otherwise specified and 1 case of acute myeloid leukemia with recurrent genetic abnormalities showed >5% p53-positive nuclei. Details regarding these cases are provided in Table 2 and representative images of a subset of these cases are provided in Figure 1 (images a, n, and o). Thus, while an elevated proportion of p53-positive cells is found more frequently in acute myeloid leukemia with myelodysplasia-related changes (Table 1), our data does not support the notion that positive p53 immunohistochemistry can be used as a surrogate for acute myeloid leukemia with myelodysplasia-related changes.

Table 2 Five cases of acute myeloid leukemia–not otherwise specified (AML-NOS) and one case of acute myeloid leukemia with recurrent genetic abnormalities (AML-RGA) with elevated p53 expression
Full size table

High p53 Expression Correlates With TP53 Mutation

Evaluation of the 67 cases for which TP53 genotype data was available showed that cases with higher p53 expression appeared to be enriched for cases with non-synonymous mutations in TP53 (Figure 3a). There were two cases with TP53 mutations in which the proportion of p53-positive cells was <5%. One case had a frameshift mutation and another case had an R196* nonsense mutation (representative image of the latter case provided in Figure 1l). In examining these two cases, we could not identify a staining pattern that made these cases clearly discernible from the numerous wild-type TP53 cases with <5% p53-positive cells.

Figure 3
figure 3

Distribution of p53 expression among cases for which TP53 was genotyped (a, n=67) and for which karyotype data was available (b, n=130). In (a) and (b), the * highlights one case of acute myeloid leukemia–not otherwise specified with normal karyotype from a patient who was alive at 423 days of follow up. WT, wild type. NA indicates cases for which TP53 was not sequenced (a) and cases for which karyotype was not available (b).

Full size image

Using the Fisher’s exact test, a cutoff of >5% was significantly associated with TP53 mutation (P<0.0001). In order to evaluate the performance characteristics of utilizing p53 expression as a predictor for TP53 mutation, we constructed a receiver operating characteristic curve using different p53 expression levels and the results are shown in Table 3. The area under the curve when using p53 expression to detect TP53 mutations was 0.873 (95% CI=0.745–1.000). Using a cutoff of >5% p53-positive cells resulted in a sensitivity of 83% and a specificity of 89% for detecting a TP53 mutation. Using a cutoff of >10% reduced the sensitivity to 0.75 and yielded an increase in specificity to 91%.

Table 3 Results of receiver operating characteristic curve for p53 expression and TP53 mutation
Full size table

As most of the cases that express p53 at >5% in our study were acute myeloid leukemia with myelodysplasia-related changes, we performed a receiver operating characteristic curve analysis to evaluate the utility of using p53 expression for predicting a TP53 mutation in this subset of acute myeloid leukemias. As shown in Table 3, the area under the curve (0.844 95% CI=0.689–0.999) was similar to that seen when including all acute myeloid leukemia subtypes. The sensitivities and specificities using >5, >10, and >15% cutoffs were also similar to those calculated when including all cases for which TP53 was genotyped. We conclude that p53 expression as determined by immunohistochemistry accurately predicts TP53 mutation status in acute myeloid leukemia when more than 5% of cells are positive.

Association Between p53 Expression and Complex Karyotype

In our cohort, information regarding karyotype was available for 130 cases, of which 102 (78%) had a normal karyotype and 28 (22%) had a complex karyotype. Of these 130 cases, 63 (48%) were acute myeloid leukemia with myelodysplasia-related changes, 36 (28%) were acute myeloid leukemia–not otherwise specified, 24 (18%) were acute myeloid leukemia with recurrent genetic abnormalities, and 7 (5%) were therapy-related acute myeloid leukemia. To evaluate the association between p53 expression and karyotype, we examined the distribution of p53 expression and observed that cases with higher levels of p53 expression appeared to consist preferentially of cases with complex karyotype (Figure 3b). A cutoff of >5% p53-positive cells was significantly associated with a complex karyotype (P<0.0001, Fisher’s exact test). To evaluate the sensitivity and specificity of p53 immunohistochemistry for detecting cases with complex karyotype, we created an receiver operating characteristic curve using the two variables. Using all subtypes of acute myeloid leukemia, the calculated area under the curve was 0.761 (95% CI=0.642–0.880) and cutoffs of >5, >10, and >15% resulted in specificities of >90% but low sensitivities for detecting cases with complex karyotype (Table 4). Restricting the analysis to cases of acute myeloid leukemia with myelodysplasia-related changes, and using the same p53 expression cutoffs, similarly showed specificities of >90% but low sensitivities. We conclude that expression p53 in >5% of cells in a bone marrow core biopsy involved by acute myeloid leukemia is not particularly sensitive but is highly specific for a complex karyotype.

Table 4 Results of receiver operating characteristic curve for p53 expression and karyotype
Full size table

p53 Expression in Cases With Abnormalities of Chromosome 17

As the TP53 gene resides on the short arm of chromosome 17 (17p13),17 we decided to evaluate whether p53 expression was related to cytogenetic abnormalities that included this region. Eleven of the 130 that were evaluated for cytogenetic abnormalities had -17 on karyotype. Of these 11 cases, 10 cases were acute myeloid leukemia with myelodysplasia-related changes and 1 case was therapy-related acute myeloid leukemia and 5 of the 11 cases had >5% p53-positive cells. TP53 was sequenced in 7 of the 11 cases, and 5 of these had a mutation in the gene, consistent with previous observations in other malignancies in which loss of one TP53 gene is often associated with mutation in the remain TP53 allele.18 The 2 cases that had a −17 and were TP53 WT were both acute myeloid leukemia with myelodysplasia-related changes with ≤5% p53-positive cells.

p53 Expression is an Independent Poor Prognostic Factor

In order to determine whether p53 expression correlates with prognosis, we obtained progression-free survival and overall survival for 115 cases for which follow-up data were available. Evaluation of different cutoffs of p53 expression showed that there was no effect on progression free survival but that overall survival was significantly shorter in cases with >5% p53 expression when examining all subcategories together as well as when restricting the analysis to cases of acute myeloid leukemia with myelodysplasia-related changes or simply acute myeloid leukemia with myelodysplasia-related changes cases with complex karyotype (Figure 4). We tested the effect of different cutoffs of p53 expression on overall survival and found that cutoffs between >5% and >20% stratified all acute myeloid leukemia cases into prognostically significant groups (Supplementary Table 1). These same cutoffs remained significantly associated with overall survival when limiting the survival analysis to the acute myeloid leukemia with myelodysplasia-related changes group (Supplementary Table 2). Within the group of acute myeloid leukemia with myelodysplasia-related changes with complex karyotype, cutoffs of >5 to >15% stratified these cases into prognostically significant groups (Supplementary Table 3).

Figure 4
figure 4

Overall survival in cases with ≤15% p53 expression compared with >15% p53 expression in all cases of acute myeloid leukemia (a, n=115), cases of acute myeloid leukemia with myelodysplasia-related changes (b, n=61), and acute myeloid leukemia- myelodysplasia-related changes cases with complex karyotype (c, n=21). Kaplan–Meier curves and statistical tests were performed using IBM SPSS software. AML-MRC, acute myeloid leukemia with myelodysplasia-related changes.

Full size image

To determine if the prognostic impact of p53 expression was independent of the factors of age, karyotype, and TP53 genotype, we initially performed univariate regression analysis (Table 5). This analysis showed that age, karyotype, and p53 score (using a cutoff of >15%) were significantly associated with survival. The lack of an association with survival with TP53 mutational status in our analysis is likely because only 47 of the cases that were genotyped for TP53 had data regarding overall survival. Subsequent multivariate analysis using age, karyotype, and p53 expression cutoff of >15% showed that age and p53 expression were independently significantly associated with survival (Table 6).

Table 5 Univariate analyses for overall survival in all cases with clinical follow up
Full size table
Table 6 Multivariate Cox proportional hazard model for overall survival in all cases with clinical follow up
Full size table

A similar analysis using the variables of age, karyotype and p53 expression performed with only acute myeloid leukemia with myelodysplasia-related changes cases showed that age and p53 expression were associated with hazard of death in univariate analysis (Table 7). In a multivariate analysis using age and p53 expression, a p53 expression level of >15% was associated with a hazard ratio (HR) for death of 4.7 (95% CI 2.08–10.6; P=0.0002; Table 8). Age was also independently associated with death (HR=1.03, 95% CI 1.003–1.06, P=0.033). As seen in the analysis performed using all cases, in the acute myeloid leukemia with myelodysplasia-related changes subgroup, the presence of TP53 mutation was not independently associated with an increased risk of death. The discrepancy between the association between TP53 mutation and p53 expression may be due to the fewer number of cases for which TP53 mutation and survival data were available (n=46, with 20 events) compared with the number of cases for which data regarding p53 expression and survival data were available (n=105, with 48 events).

Table 7 Univariate analyses for overall survival in acute myeloid leukemia with myelodysplasia-related changes (AML-MRC) subgroup
Full size table
Table 8 Multivariate analyses for overall survival in acute myeloid leukemia with myelodysplasia-related changes (AML-MRC) subgroup
Full size table

Discussion

Our findings largely agree with previous work showing that p53 is overexpressed in myeloid neoplasms with nonsynonymous TP53 mutations.19, 20 Lepelley et al12 reported that 2 of 19 cases of acute myeloid leukemia demonstrated detectable p53 expression by immunohistochemistry and these cases also had missense TP53 mutations. In this study two other cases with no p53-positive cells had a nonsense mutation or a frameshift mutation in TP53. However, this study preceded subclassification of acute myeloid leukemia into more modern prognostically relevant subtypes.12 A later study by Zolota and colleagues evaluated p53 expression in 42 bone marrow biopsies classified according to FAB classification.13 In contrast to our findings, in their study, they found a higher proportion of cases with an elevated fraction of p53-postive nuclei. Specifically, 81% of the cases in their cohort showed ≥10% p53-positive nuclei. We speculate that this difference is due to the higher titer of DO-7 antibody used in their study (1:150) compared with the titer used in our study (1:400).

In contrast to the prior studies, our cohort is subclassified according to the 2016 WHO subgroups. Our study also included a relatively high percentage of acute myeloid leukemia with myelodysplasia-related changes, possibly due to the predominance of older adults in the study, which is a particularly aggressive subtype of acute myeloid leukemia. To our knowledge, our study is the largest cohort of acute myeloid leukemia with myelodysplasia-related changes to be analyzed for p53 expression and correlated with TP53 mutation and survival. Another study evaluated p53 expression in acute myeloid leukemia evolving from a preceding myelodysplastic syndrome.15 In this report, 9 of 39 cases (23%) of secondary acute myeloid leukemia showed strong nuclear staining (score 3+) in at least 5% of the cells in the entire bone marrow. Their scoring strategy differed from ours in that they only counted strongly positive cells. The authors sequenced TP53 in 37 randomly selected bone marrow samples from their cohort of higher-risk myelodysplastic syndrome and secondary acute myeloid leukemia and found substantial agreement with p53 (Cohens Kappa=0.645). In lower risk myelodysplastic syndrome with del(5q)19 and in therapy-related myeloid neoplasms,20 a p53 cutoff of ≥1% was found to independently predict outcomes. Overall, there is likely inter-laboratory variability in terms of p53 expression and thus any published cutoff for interpretation of immunohistochemistry expression should be validated with cases known to have abnormal TP53. Furthermore, it is probably prudent to evaluate TP53 genotype in cases with elevated p53 expression.

Potential limitations of our study include our use of the proportion of nucleated cells expressing p53 in the marrow rather than the proportion of blasts expressing p53. However, it is likely that at least a subset of the non-blast cells in marrows with leukemia are derived from the neoplastic clone, have TP53 mutations, and consequently exhibit a p53-mutant immunophenotype. This proposal is supported by the cases in our cohort that had megakaryocytes with strong nuclear p53 staining and we are not the first group to report strong p53 expression in non-blast cells in cases of myelodysplastic syndrome with TP53 mutations.19, 20 In addition, accurate quantification of the fraction of blasts expressing p53 would be challenging and would likely require staining with additional markers to accurately quantify blasts in the marrow. Thus, we do not see assessment of p53 expression as a proportion of all nucleated cells as a limitation. In any case, our study demonstrates that determining p53 expression as a fraction of all nucleated cells in the marrow is prognostically useful, predicts TP53 mutation, and is likely easily incorporated into clinical practice. It is also important to note, that, because our cohort contains relatively few cases with a high level of p53 expression and because we did not test cutoffs with a validation set, the cutoffs used here should be used as general guideline rather than as definitive values.

An additional limitation of our study is the approach used to sequence the TP53 gene. First, we did not sequence the entire TP53 gene, and thus there are possibly cases in which TP53 is mutated that were called TP53 WT in our study. As our sequencing targets known hotspots, the number of cases with mutations that were called WT is likely small, if indeed there are any. Despite these limitations, which would potentially result in an underestimate of the sensitivity and specificity of p53 immunohistochemistry to predict mutation status, our study suggests that p53 immunohistochemistry has a reasonable sensitivity and specificity for detecting TP53 mutations. In addition, the sensitivity and specificity of p53 immunohistochemistry for identifying TP53-mutated cases determined in our study are similar that seen in lung and ovarian tumors,17, 21 suggesting that at least some of the limitations facing the use of immunohistochemistry in predicting TP53 abnormalities are shared across diverse neoplasms.

As sequencing of TP53 becomes routinely incorporated into sequencing panels used in the diagnostic workup of acute myeloid leukemia, it is reasonable to ask whether p53 immunohistochemistry would be redundant and unnecessary in most cases. Although our data suggest that p53 immunohistochemistry may be a useful ancillary test in cases of acute myeloid leukemia with myelodysplasia-related changes to identify patients with a particularly poor prognosis, it is probably unnecessary to perform immunohistochemistry for p53 in all cases of acute myeloid leukemia. Because the turnaround time for immunohistochemistry is typically much shorter than molecular assays to assess TP53 genotype, it may be useful to perform p53 immunohistochemistry when a bone marrow biopsy shows features that are suspicious for dysplasia or in the case of patients with acute myeloid leukemia evolving from a preceding myelodysplastic syndrome. In addition, there are situations in which cytogenetic data is unavailable (eg, due to technical limitations, insufficient material, absent analyzable metaphases, etc) and in these cases, p53 immunohistochemistry may be useful to suggest whether a case of acute myeloid leukemia will be particularly aggressive.