Cell cycle regulators in multiple myeloma: Prognostic implications of p53 nuclear accumulation☆

Patients and samples 

The study population includes 48 consecutive MM patients admitted to the Hematology Service, Ospedale Maggiore IRCCS, Milan between 1996 and 1999 and already characterized for the occurrence of the t(11;14) translocation and cyclin D1 IR.19 The diagnosis of MM was based on the Durie and Salmon criteria.20 Bone marrow aspirates and biopsies were simultaneously taken from each patient and examined by conventional light microscopy and phenotypical analysis. There were 40 (83.3%) patients at diagnosis and 8 (16.7%) relapses; 22 patients (45.8%) were males and 26 (54.2%) females, with a median age of 65 years (range, 36 to 85 years). The serum monoclonal component was IgG in 26 cases (54.2%), IgA in 16 cases (33.3%), and IgD in 1 case (2.1%); 4 cases (8.3%) were λ light chain MMs, and 1 case (2.1%) was IgA/IgG. The serum monoclonal component κ/λ chain ratio was 1.4:1. The Durie and Salmon stage and the record of clinical symptoms at diagnosis were available in all cases. Bartl's histologic grade and stage21 were assessed in the 40 patients at diagnosis. Clinical follow-up was available for 46 patients, with a median of 36 months (range, 2 to 74 months). Follow-up of the 8 patients analyzed at relapse was referred to the first diagnosis. Twenty patients (43.5%) died of disease, and 26 (56.5%) were alive with disease. Thirty-four patients (73.9%) were treated with chemotherapy because of symptomatic disease (ie, bone pain, anemia, fever), 22 at diagnosis and 12 after a median time of 15 months (range, 10 to 48 months). Twenty-six patients were frontline treated with conventional melphalan/prednisone combination, 8 with high-dose melphalan with tandem transplants of autologous peripheral blood stem cells.22 Control experiments were performed on normal bone marrow biopsies from 10 patients with localized solid tumors. All of the 58 bone marrow biopsies (including control tissues) were fixed in B5 and decalcified by EDTA (Mielodec; Bio-Optica, Milan, Italy). The mean length of the biopsy samples was 15 mm (range, 7 to 28 mm). Informed consent was obtained from all patients investigated.

Immunohistochemistry 

IR for p53, p21, HDM-2, p27, cyclin E, and Ki-67 was evaluated by the avidin-biotin peroxidase complex (ABC) method using the 3,3'-diaminobenzidine tetrahydrochloride chromogen as previously described.19 For antigen retrieval, the slides were placed in 0.1 M citrate buffer, pH 6.0 (for p53, p21, p27, and Ki-67), 0.01 M EDTA buffer, pH 8.0 (for cyclin E), or 0.1 M bicarbonate buffer, pH 6.1 (for HDM-2), and underwent 3 (for p53, p21, p27, and Ki-67), 4 (for cyclin E), or 6 (for HDM-2) 5-minute cycles at 90°C in a 780-W microwave oven. The sections were immunostained with the polyclonal antibody CM1 to p53, which recognizes either the wild-type mutant p53 protein (Novocastra, Newcastle upon Tyne, U.K.), at a working dilution of 1:3000), the WAF-1 monoclonal antibody (MoAb) to p21 (dilution 1:50; Oncogene, New York, NY), the IF-2 MoAb to HDM-2 (1:20; Oncogene), the Kip1 MoAb to p27 (1:200; Transduction Laboratories, Lexington, U.K.), the 13A3 MoAb to cyclin E (1:50; Novocastra) and the Mib-1 MoAb to Ki-67 (1:100; Immunotech, Marseille, France).

At least 500 neoplastic plasma cells were evaluated at ×1000 magnification without knowledge of the genetic and clinicopathologic data, and the percentage of cells showing nuclear staining was recorded. According to the commonly adopted cutoff criteria,12, 17 only cases with ≥10% (for p53, p21, HDM-2, cyclin E, and Ki-67) and ≥50% (for p27) immunoreactive cells were considered positive. To assess the nature of immunoreactive cells in cases with an interstitial infiltration, double-immunostaining experiments with the 5F7 MoAb to the anti-plasma cell–associated antigen CD138 (1:50; Novocastra) were also performed. Formalin-fixed, paraffin-embedded samples of laryngeal squamous cell carcinomas (for p53, p21, HDM-2, p27, and Ki-67) and of invasive duct carcinomas of the breast (for cyclin E) served as positive controls. Negative control sections were incubated with nonimmune rabbit or mouse serum in place of the specific primary Abs.

Polymerase chain reaction–single-strand conformation polymorphism and southern blot analyses 

Mutations in exons 5 through 8 of TP53 gene and HDM-2 gene amplification were investigated in bone marrow aspirates obtained simultaneously with bone marrow biopsy by polymerase chain reaction–single-strand conformation polymorphism (PCR-SSCP) and direct sequencing analyses and by Southern blot and quantitation densitometric analyses, respectively.5, 23 Exons 5 to 8 of the p53 gene were amplified separately using specific oligonucleotide primers. Briefly, PCR was performed using 100 ng of genomic DNA, 10 mmol Tris-HCl (pH 8.8), 50 mmol KCl, 1mmol MgCl2, 100 μM of each of the deoxyribonucleoside triphosphates, 0.25 U Taq DNA polymerase, and 10 pmol of each primer, in a final volume of 25 μL. Thirty cycles of denaturation (15 seconds at 94°C), annealing (1 minute at 65°C for exons 5, 6, or 7 and at 58°C for exon 8), and extension (1 minute at 72°C) were performed. Ten μL of the reaction mixture was mixed with an equal volume of a denaturing stop solution. Samples were heated at 95°C for 5 minutes, chilled on ice, and immediately loaded onto a 6% acrylamide–Tris-borate-EDTA gel. The samples were run at 6 to 8 W constant power for 14 hours at room temperature. Detection was performed by the silver staining method. PCR products showing an aberrant migration pattern were directly sequenced in both directions using the 5' and 3' primers used for the PCR-SSCP analysis. The DNA fragments were purified by agarose gel extraction and sequenced using the Big Dye Terminator Sequencing Kit in an ABI Prism 310 automated sequencer (PerkinElmer Cetus, Norwalk, CT). The HDM-2 gene was investigated using a 2.4-Kb probe corresponding to the HDM-2 cDNA. To assess the degree of amplification, a quantitative densitometric analysis of at least 2 restriction DNA digests from each sample was made using a GS-670 Image Densitometer (Bio-Rad, Richmond, CA). A 6.6-Kb BamHI-HindIII fragment representative of the joining (JH) region of the immunoglobulin heavy chain locus was chosen as a suitable single-copy gene probe for normalization of the DNA content. The percentage of neoplastic plasma cells in bone marrow aspirates, as evaluated by flow cytometric analyses after Ficoll gradient separation using the anti-CD138 antibody ranged from 17% to 77% (mean, 33%).

Statistical analysis 

The correlations among p53, p21, HDM-2, cyclin E, cyclin D1, and Ki-67 IR and the clinicopathologic and immunohistochemical data were evaluated by Fisher's exact test. The following variables were investigated for their influence on overall survival: gender (male and female), age (≤60 years and >60 years), lactic dehydrogenase level (≤460 and >460 U/L), β2 microglobulin (≤2.6 and >2.6 μg/dL), clinical stage (I, II, and III), clinical symptoms (presence and absence), Bartl's histologic grade (low-, intermediate-, and high-grade malignancy), Bartl's histologic stage (≤19%, 20% to 50%, and >50% of bone marrow infiltration), and immunoreactivity for p53, p21, HDM-2, p27, and cyclins E and D1 (positive and negative). Furthermore, to evaluate the survival distribution trend, p53 IR was analyzed also by grouping patients in 3 categories, those with unreactive tumors and those with tumors with 1% to 9% and ≥10% p53 IR.

Survival estimates were calculated using the Kaplan-Meier method and compared using the log-rank test. The Cox proportional hazard regression model was used to evaluate the simultaneous effect of explanatory variables on survival time. Statistical analysis was made using SAS software (Statistical Package, version 6.12, SAS Institute, Cary, NC).

Immunohistochemistry and molecular analyses 

Nonneoplastic mononuclear cells of the bone marrow were unreactive for p53. Cyclin E was detectable in a minor subset of erythroid and myeloid precursor cells, and p27 IR was a consistent feature of most megakaryocytes, endothelial cells, plasma cells, and lymphocytes. Ki-67 was detectable in a significant fraction of granulocytic and erythroid precursors and in megakaryocytes. IR for p21 and HDM-2 was restricted to megakaryocytes (data not shown).

Nineteen (39.6%) cases had ≥10% p53 immunoreactive tumor cells (Fig 1A).

View Large Image Download to PowerPoint

Fig. 1. Examples of the immunohistochemical results. Nuclear immunoreactivity (IR) for (A) p53, (B) p21, and (C) HDM-2 in plasma cells of different cases of MM. (D) Double immunostainings for p27/CD138 and (E) cyclin E/CD138 showing diffuse IR for p27 and cyclin E (in brown) in most of the CD138-positive (in blue) neoplastic plasma cells. (F) A case of MM with a high Ki-67 labeling index. (Hematoxylin counterstain [A, B, C, and F]; original magnification × 40.)

Twenty-eight (58.3%) tumors had ≥10% HDM-2 immunoreactive tumor cells (Fig 1C). None of the cases showed HDM-2 gene amplification according to our Southern blot and densitometric assay.

Twenty-three (47.9%) tumors had ≥10% neoplastic cells immunoreactive for p21, 29 (60.4%) for cyclin E, and 16 (33.3%) for Ki-67. Twenty-three (47.9%) tumors had ≥50% neoplastic cells immunoreactive for p27 (Fig 1B, D, E, and F).

Correlations among p53, p21, p27, HDM-2, cyclin E, and cyclin D1 IR 

p21 and HDM-2 IR in ≥10% of tumor cells were detected in 3 and 2, respectively, of the 3 cases with the mutated TP53 gene. No correlation between p53 accumulation and HDM-2 (P = 0.258) or p21 (P = 0.285) IR was found in the 31 tumors with the wild-type TP53 gene. Among the other cell cycle–related proteins analyzed, there was a trend for a direct association between p27 and cyclin E IR, which approached statistical significance (P = 0.083). Finally, a marginally significant direct association was found between Ki-67 labeling and cyclin E IR (P = 0.073).

Twelve (25%) of the cases were previously reported to express cyclin D1 protein,19 and this showed a significant inverse correlation with p27 (P = 0.002) and cyclin E (P < 0.001) IR. In particular, IR for p27 in ≥50% of neoplastic cells was detected in 1 (8.3%) of the 12 cyclin D1–positive cases and in 22 (61.1%) of the 36 cyclin D1–negative cases, and IR for cyclin E in ≥10% of neoplastic cells was detected in 2 (16.6%) cyclin D1–positive and in 27 (75%) of the cyclin D1–negative cases.

Clinicopathologic correlates 

Cyclin E IR was directly associated with poorly differentiated tumors (P = 0.018). A significant (P = 0.005) direct correlation was found between p53 IR and the degree of bone marrow infiltration. Finally, cyclin E IR was inversely associated with the tumor burden at a borderline level of statistical significance (P = 0.083). No association was found among p53, p21, HDM-2, p27, and cyclin E IR and the clinicopathologic characteristics of age, gender, clinical stage, β2-microglobulin, and lactic dehydrogenase (data not shown).

Survival analyses 

Univariate analysis of survival showed that clinical stage (P < 0.0001), Bartl's histologic stage (P = 0.016), and grade (P = 0.027) were significantly associated with survival. The association between survival and the variables of age and gender was of borderline significance (P = 0.091 and P = 0.061, respectively). No prognostic implications were ascertained for p21, HDM-2, p27, cyclin E, cyclin D1, and Ki-67 IR (Table 1).

Table 1. Univariate analysis for the 46 patients with available follow-up

		Subjects (n = 46)	Deaths (n = 20)	P value (log-rank)
	Age
	≤60	20	4
	>60	26	16	0.091
	Gender
	Males	20	10
	Females	26	10	0.061
	Clinical stage
	I	25	3
	II	7	4
	III	14	13	<0.0001
	Bartl's stage
	I	20	4
	II	11	5
	III	8	5
	Relapses	7	6	0.016
	Bartl's grade
	I	29	9
	II	8	3
	III	2	2
	Relapses	7	6	0.027
	β2 microglobulin
	0	15	5
	1	31	15	0.459
	LDH
	0	42	19
	1	4	1	0.992
	p53
	<10%	29	10
	≥10%	17	10	0.029
	p53
	0%	12	5
	1–9%	17	5
	≥10%	17	10	0.063
	HDM-2
	<10%	19	7
	≥10%	27	13	0.972
	p21
	<10%	23	10
	≥10%	23	10	0.435
	p27
	<50%	25	11
	≥50%	21	9	0.822
	Cyclin E
	<10%	19	10
	≥10%	27	10	0.282
	Cyclin D1
	<10%	35	14
	≥10%	11	6	0.346
	Ki-67
	<10%	29	14
	≥10%	17	6	0.649

NOTE. LDH: O, ≤460 U/L; 1, >460 U/L. β2 microglobulin: 0, ≤2.6 μg/dL; 1 >2.6 μg/dL. Bartl's grade: I, low; II, intermediate; III, high-grade malignancy. Bartl's stage: I, <20%; II, 20% to 50%; III, >50% of bone marrow infiltration.

In univariate analysis, p53 accumulation was significantly associated with survival in the whole series (P = 0.029) (Fig 2 and Table 1), as well as in the 39 patients at diagnosis and available follow-up (P = 0.055) or the 34 patients treated by chemotherapy (P = 0.038).

View Large Image Download to PowerPoint

Fig. 2. Five-year overall survival of the 46 patients with available follow-up according to IR for p53.

Table 2. Multivariate analysis for the 46 patients with available follow-up

		HR (95% CI)	P value	P for trend
	p53
	<10%	1.00
	≥10%	4.20 (1.22–14.4)	0.022
	p53
	0%	1.00
	1–9%	3.04 (0.43–21.3)	0.263
	≥10%	7.24 (1.39–37.8)	0.018	0.017

NOTE. Hazard's ratio (HR) and 95% confidence interval (CI) are adjusted for age, stage, and relapse.