Heat shock proteins HSP27, HSP60, HSP70, and HSP90

Expression in bladder carcinoma


  • Thierry Lebret M.D., Ph.D.,

    1. Department of Urology and Pathology, Hôpital Foch, Suresnes, France
    2. Department of Surgery, University College Dublin, Mater Misericordiae University Hospital, Conway Institute of Biomolecular and Biomedical Research, Dublin, Ireland
    Search for more papers by this author
  • R. William G. Watson Ph.D., B.Sc.,

    Corresponding author
    1. Department of Surgery, University College Dublin, Mater Misericordiae University Hospital, Conway Institute of Biomolecular and Biomedical Research, Dublin, Ireland
    • Department of Surgery, Mater Misericordiae University Hospital, 47 Eccles Street, Dublin 7, Ireland
    Search for more papers by this author
    • Fax: (011) 353-1-8300345

  • Vincent Molinié M.D.,

    1. Department of Urology and Pathology, Hôpital Foch, Suresnes, France
    Search for more papers by this author
  • Amanda O'Neill Ph.D,

    1. Department of Surgery, University College Dublin, Mater Misericordiae University Hospital, Conway Institute of Biomolecular and Biomedical Research, Dublin, Ireland
    Search for more papers by this author
  • Christophe Gabriel M.D.,

    1. Department of Urology and Pathology, Hôpital Foch, Suresnes, France
    Search for more papers by this author
  • John M. Fitzpatrick M.Ch.,

    1. Department of Surgery, University College Dublin, Mater Misericordiae University Hospital, Conway Institute of Biomolecular and Biomedical Research, Dublin, Ireland
    Search for more papers by this author
  • Henry Botto M.D.

    1. Department of Urology and Pathology, Hôpital Foch, Suresnes, France
    Search for more papers by this author



Heat shock proteins (HSPs) are synthesized by cells in response to various stress conditions, including carcinogenesis. The expression of HSPs in neoplasia has been implicated in the regulation of apoptosis, and HSPs also can act by increasing immunity. In the current study, the authors attempted to clarify the significance of HSPs in bladder carcinoma and their effect on tumor behavior.


Expression levels of the 27-kilodalton HSP (HSP27), HSP60, HSP70, and HSP90 were studied using immunohistochemistry on tissue sections from 42 transitional cell carcinomas of the bladder (14 Grade 1 tumors; 13 Grade 2 tumors; 15 Grade 3 tumors, including 3 tumors associated with carcinoma in situ; 30 Stage Ta tumors; 7 Stage T1 tumors; and 5 Stage T2 tumors). Bladder specimens from 10 healthy patients were used as controls in the study. The selected patients had a mean follow-up of 52 months (range, 24–78 months). Among the 37 patients with superficial bladder carcinoma, 17 patients did not have any recurrence after undergoing primary resection, and 20 patients developed recurrent disease, including 4 recurrences with muscle invasion. HSP expression was evaluated according to the percentage of positively stained cells, and loss of expression was defined as < 80% of stained cells.


In normal bladder specimens, all four HSPs (HSP27, HSP60, HSP70, and HSP90) were expressed strongly in the cytoplasm and membrane from the basal cell layer to the superficial cell layer. Loss of expression was detected in tumors: respectively, 45.2%, 38.1%, 69.0% and 23.8% of tumors showed a loss of immunostaining for HSP27, HSP60, HSP70, and HSP90. No correlation between HSP expression and grade was found. Low expression levels of HSP27 and HSP60 were correlated with higher tumor stage (87% vs. 6% [P < 0.001] and 78% vs. 9% [P < 0.01], respectively). HSP60 and HSP90 expression levels were correlated with final outcome for patients with superficial bladder carcinoma: loss of expression was associated with the risk of developing an infiltrating recurrence (97% vs. 6.0% [P < 0.001] and 88.2% vs. 52.5% [P = 0.02] for HSP60 and HSP90, respectively).


HSPs were expressed in normal urothelium, and the current results indicated that loss of HSP60 and HSP90 expression may have prognostic relevance in patients with bladder carcinoma. The authors believe that HSP60 may be a very useful marker for patients with superficial bladder carcinoma and may be used for predicting disease progression. If these data are confirmed, low HSP60 expression levels may be usable as a prognostic marker to identify patients for whom local treatment would be insufficient. Cancer 2003;98:970–7. © 2003 American Cancer Society.

DOI 10.1002/cncr.11594

Bladder carcinoma represents the fourth most common malignancy in males and causes about 3.5% of all deaths from malignant disease.1 Nevertheless, most tumors (70–80%) are diagnosed as superficial (no muscle invasion: < T2), but approximately 70% of these tumors will recur. These tumors progress either in stage or grade, and 10–15% advance to muscle invasion and metastases.2, 3 The behavior of the initial tumorigenesis cannot always be predicted accurately, because the molecular pathogenesis of bladder tumor remains poorly understood. Genetic alterations, including mutations in p53, p21, or retinoblastoma genes and over-expression of Bcl-2, c-myc, or c-erb B2, have been studied extensively, but not all molecular alterations have been identified in transitional cell carcinoma (TCC).4, 5 Understanding the molecular mechanisms involved in TCC of the bladder may lead to improvements in patient management.

Resistance of tumor cells to apoptosis enhances their spontaneous growth and renders them resistant to host defense mechanisms as well as to various forms of therapy. Heat shock proteins (HSP) are molecular chaperones with a number of functions. They are a highly conserved group of proteins generated in all organisms in response to heat as well as many other forms of stress, including hypoxia, irradiation, infection or toxic agents.6 Their functions are ubiquitous, and in neoplasms, the expression of HSP has been implicated in the regulation of apoptosis, as a modulator of p53 function, in the immune response against tumors, and in multidrug resistance.7, 8 Increased HSP levels make cells more resistant to apoptosis but can also act by increasing their immunity, because these proteins are among the most immunogenic molecules known.9 It is not surprising, therefore, that, during carcinogenesis, HSP expression is altered in many tumors.10 HSP expression levels in bladder tumors have not been studied extensively. Understanding the specific roles of HSPs in urothelial carcinogenesis has important implications regarding tumor behavior and potential prognostic implications.11

The objective of the current study was to clarify the significance and the usefulness of HSPs in bladder carcinoma. We hypothesized that HSP may play a prognostic or diagnostic role in patients with primary bladder tumors. Therefore, samples of TCC of the bladder were obtained from patients undergoing transurethral tumor resection or cystectomy and were examined for HSP expression. Expression levels were compared with levels measured during clinical follow-up. HSPs have been classified according to molecular weight; HSPs include the 27-kilodalton (kD) HSP (HSP27), HSP60, HSP70, and HSP90, the 4 HSPs examined in the current study.



After obtaining ethical approval from our French institutional review committee and the French Ligue Contre le Cancer, archival surgical bladder specimens were obtained from 52 consecutive patients (11 female patients and 41 male patients) between January 1995 and June 1995; 42 patients had primary TCC of the bladder, and 10 were normal-bladder control patients. During this period, all patients who had undergone a tumor bladder resection (n = 37) or a cystectomy (n = 5) for TCC of the bladder were included. Normal bladder specimens came from transuretral resection of the prostate (healthy bladder mucosa). Patients with invasive tumors (> T1) underwent radical cystectomy, and none of those patients had tumors associated with positive lymph nodes. The tissues were fixed either in formaldehyde or in Bouin solution, embedded in paraffin, then histologically evaluated in sections stained in hematoxylin and eosin.


The hematoxylin and eosin–stained sections of all specimens were reviewed by one pathologist (V.M.) and classified according to the Ancona 2001 refinement of the World Health Organization 1973 classification12 and the 1997 pTNM pathologic staging system13 (Table 1).

Table 1. Status and Grade of 42 Bladder Tumorsa
GradeTumor statusTotal no.
TaT1> T1
  • a

    Forty-two bladder tumors were studied: 37 superficial tumors from patients who underwent transurethral resection and 5 invasive tumors from patients who underwent cystectomy.

  • b

    One tumor was associated with carcinoma in situ.

  • c

    Two tumors were associated with carcinoma in situ.



Three-micrometer sections of tissue were placed on poly-L-lysine coated slides, deparaffinized in xylene, and rehydrated in alcohol. Sections then were incubated with 0.3% H2O2 in methanol to inhibit endogenous peroxidase. For epitope retrieval before immunostaining, slides were pretreated with 0.01 M citrate buffer, pH 6.0, for 20 minutes in a microwave (power, 750 watts). The slides were incubated at room temperature for 2 hours in a humidified chamber with mouse monoclonal antibodies to anti-HSP27 (1:200 dilution), anti-HSP60 (1:500 dilution), anti-HSP70 (1:300 dilution; from StessGen Biotechnologies Corp., Victoria, British Columbia, Canada), and anti-HSP90 (1:200 dilution; from Novocastra Laboratories Ltd., Newcastle upon Tyne, UK).

The standard biotin-avidin system (Vectasion ABC Kit; Vector Laboratories, Burlingame, CA) was used for antibody localization, which was performed according to the vendor's protocol. For all primary antibodies, the secondary antibodies used were mouse monoclonal antibodies.

Sections subsequently were washed, developed with 3,3′-diaminobenzidine-hydrogen peroxidase, and counterstained with hematoxylin. For negative control sections, the primary antibody was replaced by normal swine serum; and for positive control sections, pleomorphic adenoma of the salivary gland was used.14

Evaluation HSP Expression

The extent of immunostaining was determined by one blinded, independent, experienced uropathologist (V.M.) who had no knowledge of patient history or clinical behavior. HSP expression was evaluated according to the proportion of positively stained cells and their exact localization. For each tumor, the percentage of immunostaining cells was estimated per 1000 cells. Loss of expression was defined as < 80% stained cells.

Clinical Follow-Up

All patients had a minimum follow-up of 24 months (mean, 52 months; range, 24–78 months). According to European Association of Urology guidelines, patients with Stage T1A, Grade 1–2 tumors did not receive any adjuvant therapy; whereas for patients with superficial Grade 3 tumors or carcinoma in situ, 6 weekly, standard 81 mg bacillus Calmette–Guerin (BCG) doses were administered intravesically. Among patients who had superficial bladder tumors, 3 populations were identified: Group 1 included 17 patients who did not have any disease recurrence, Group 2 included 15 patients who developed recurrent disease with superficial lesions, and Group 3 included 5 patients with tumors that became invasive who underwent cystectomy.

Statistical Analysis

The data were analyzed using the StatView software program (Version 5; SAS Institute, Cary, NC). The Student t test was used to assess the significance of results (P < 0.05). A one-way analysis of variance was used to determine the correlation between protein expression and grade of tumor differentiation, pathologic stage of tumor, and follow-up. HSP expression was analyzed as a continuous variable.


Expression of HSP

In normal bladder tissue, the four HSPs (HSP27, HSP60, HSP70, and HSP90) were expressed highly in all cells. Cytoplasm and membrane staining was detected from the suprabasal cell layer to the superficial cell layers (Fig. 1).

Figure 1.

Normal bladder tissue. Very strong expression levels of the 4 heat shock proteins ([A] the 27-kilodalton heat shock protein [HSP27], [B] HSP60, [C] HSP70, and [D] HSP90) are detected in the cytoplasm and on the cell membrane in all urothelial cells.

Concerning the tumor samples, various staining patterns were observed. Loss of expression was detected in many tumors (Table 2). Moreover, this loss of expression was observed in the superficial cell layers, whereas HSP generally stained positively in the suprabasal cell layer. HSP expression varied from patient to patient, with 45.2%, 38.1%, 69.0%, and 23.8% of tumors showing a loss of immunostaining for HSP27, HSP60, HSP70, and HSP90, respectively (Fig. 2).

Table 2. Heat Shock Protein Expression Obtained by Immunohistochemical Techniques in 42 Bladder Tumor Samplesa
  • kD: kilodalton; HSP27: the 27 kD heat shock protein; HSP60: the 60 kD heat shock protein; HSP70: the 70 kD heat shock protein; HSP90; the 90 kD heat shock protein.

  • a

    Loss of expression was detected in many tumors and may represent an important event in carcinogenesis.

No. of patients42424242
Mean expression (% stained cells)
No expression (no. of samples)4321
Immunostained cells (no. of samples)    
 > 80%23261332
Figure 2.

Various staining patterns in a specimen of Stage Ta/Grade 2 bladder cell carcinoma, showing (A) 80% expression of the 27-kilodalton heat shock protein (HSP27), (B) 100% expression of HSP60, (C) no expression of HSP70, and (D) 10% expression of HSP90 (only superficial cells are immunostained).

Correlation between HSP Expression and Clinicopathologic Findings

To understand the implications of the loss of HSP expression in carcinogenesis, the expression of HSP was analyzed for correlations with clinicopathologic features (Table 3). Table 4 shows the correlations of HSP immunostaining with clinical stage and histologic grade. Correlations of loss of expression of HSP27 (P < 0.001) and HSP60 (P < 0.01) with disease stage were established: in infiltrative tumors, the mean levels of HSP27 and HSP60 expression were 6% and 9%, respectively; in superficial tumors, the mean levels of HSP27 and HSP60 expression were 86.6% and 78.3%, respectively. No correlation with histologic grade was shown. When comparing the risk of recurrence and progression for the 37 patients who had initial superficial bladder tumors with HSP expression levels, it appears that HSP70 expression does not have any correlation with recurrence or progression.

Table 3. Correlation between Heat Shock Protein Expression and Clinicopathologic Featuresa
VariableNormal bladder tissueSuperficial tumorInvasive tumor (cystectomy)
Without recurrence (Group 1)With recurrence (Group 2)With progression (Group 3)
  • kD: kilodalton; HSP27: the 27 kD heat shock protein; HSP60: the 60 kD heat shock protein; HSP70: the 70 kD heat shock protein; HSP90: the 90 kD heat shock protein.

  • a

    Expression levels were measured according to the percentage of stained cells. Loss of HSP expression was more common in infiltrative tumors (HSP27, HSP60, and HSP70) and in primitive superficial tumors that recurred (HSP27) or progressed (HSP60 and HSP90). The five tumors that recurred with progression included two initially Ta/Grade 3 tumors, one initially T1/Grade 2 tumor, and two initially T1/Grade tumors.

No. of patients10171555
HSP27 (%)     
HSP60 (%)     
HSP70 (%)     
HSP90 (%)     
Table 4. Correlations of Heat Shock Protein Expression with Disease Stage, Grade, and Outcome
 Disease stageGradeOutcome (recurrence vs. no recurrence)Outcome (infiltration vs. no infiltration)
  • kD: kilodalton; HSP27: the 27 kD heat shock protein; NS: not significant; HSP60: the 60 kD heat shock protein; HSP70: the 70 kD heat shock protein; HSP90: the 90 kD heat shock protein.

  • a

    Ta: n = 30; T1: n = 7; > T1: n = 5.

  • b

    Grade 1: n = 14; Grade 2: n = 11; Grade 3: n = 15.

  • c

    Recurrence: n = 20; no recurrence: n = 17.

  • d

    Infiltration: n = 5; no infiltration: n = 32.

No. of patients42a42b37c37d
P value    
 HSP 27< 0.001NS0.07 (NS)NS
 HSP 60< 0.01NS0.06 (NS)< 0.001
 HSP 90NSNS0.3 (NS)< 0.02

If we consider the final behavior of all primary superficial bladder tumors and compare the 2 distinct populations—patients who are cured with local treatment (no infiltrative recurrence) and patients who need more aggressive treatment, such as cystectomy or radiochemotherapy (tumors that become muscular infiltrating)—a high correlation is established with HSP60 expression (P < 0.001) and HSP90 expression (P < 0.01) (Table 4). The first group of patients had 93.3% mean HSP60 expression and 88.7% mean HSP90 expression, and the second group had 6.0% mean HSP60 expression and 52.5% mean HSP90 expression. For HSP60 expression, a cut-off level of 40% can definitively predict the behavior of the primary superficial bladder tumor: below this threshold level, all patients with primary tumors had progression and required cystectomy; in contrast, all patients who had tumors with > 40% HSP60 expression had their disease controlled with local treatment.


HSP expression has been studied in many tumor tissues, but studies in patients with TCC of the bladder have yet to be undertaken. In the current study, we examined the expression patterns of the four major HSPs in normal bladder tissue and in TCC tissue. We demonstrated that HSP27, HSP60, HSP70, and HSP90 are expressed in the normal urothelium but that a partial or complete loss of expression occurs in some tumors. In fact, HSP expression differs depending on the type of HSP involved. The levels of expression of each type of HSP have been compared with the natural clinical outcome, and this is the first study to report that a loss of expression has a prognostic significance for two different HSPs: HSP60 and HSP90. It has been shown that expression of HSP60 and HSP90 is a predictive factor for patients with superficial bladder tumors: loss of expression indicates a poor prognosis and the risk of developing tumor infiltration on follow-up. This is a major endpoint, and confirmation of these results will mean that HSP60 and HSP90 expression may become a very promising marker for patients with noninfiltrative tumors. In fact, for these patients, the challenge is to recommend the correct treatment without delay, which depends directly on the risk of recurrence. Predictive factors in patients with bladder tumors have been studied extensively, but expression levels of neither oncogenes (ras, erb B-2) nor tumor suppressor genes (p53) nor cell cycle genes (p16 and p15) are used routinely.15 Early cystectomy or concomitant radiochemotherapy may be proposed if patients have a high risk of progression; therefore, low HSP60 expression may influence the decision. Correlation of HSP60 expression with a better prognosis has been described for patients with ovarian tumors,16 with HSP60 overexpression in the initial stages of the disease and a loss of expression during evolution. Decreased HSP60 expression has been found to be correlated with tumor aggressiveness in patients with ovarian tumors. In the current study, low HSP90 expression also was associated with a poor prognosis. In agreement with the results of the current study, Cardillo et al. previously found that 92.9% of tumors expressed HSP90. Cardillo et al. described the same decrease in expression in the superficial layer for nearly 10% of tumors; however, contrary to our results, no correlation with prognosis was found.17 This point may be explained by the fact that Cardillo et al. studied invasive tumors, which have a different behavioral pattern than superficial tumors.

Oncologic research has shown that HSPs may be crucial for tumorigenesis18, 19, and for many tumor types, such as breast, colon, prostate, and liver carcinoma, it has been shown that HSPs initially are a prerequisite for the survival of tumor cells.20–22 In these tumors, HSP inhibits apoptosis and prevents cell death. Conversely, for other tumor types, like ovarian carcinoma, osteosarcoma, and melanoma, it has been shown that HSPs are involved in immunologically mediated reactions and in autoimmunity.23, 24 Our results seem to indicate that HSP plays the same role in bladder tumorigenesis, and loss of expression may decrease the efficacy of the immune response. The way in which HSPs are involved in the immune response remains unknown, although several pathways have been demonstrated. In the current study, the majority of HSPs were expressed on the surface of the cell. This is an important site for antibody assembly and antigen presentation. In particular, HSP can make complexes with cellular proteins and, with the major histocompatibility complex Class I molecules, they elicit the response of CD8 positive T cells. HSPs derived from tumor cells also are recognized by T cells, including γ/δ T lymphocytes and natural killer cells.25, 26 The partial or total loss of expression may explain the poor prognosis due to the inability to elicit an immune response. Recently, Zlotta et al. reinforced this hypothesis and suggested that HSPs have independent roles and may represent a target for humoral antitumor responses.27 As is true for patients with melanoma, it is known that immunotherapy is one of the efficient therapies for patients with TCC of the bladder. Zlotta et al. examined HSP60 expression during the humoral response stimulated by BCG in the treatment of patients with superficial bladder carcinoma. Intravesical BCG therapy is accepted as the treatment of choice for patients with noninvasive bladder tumors who have a high risk of recurrence.28–30 BCG therapy uses the Mycobacterium bovis, which contains a very immunogenic protein: HSP60 (65 kD). Whereas bacillus adhesion to bladder tumoral cell induces the Th1 immune response with recruitment of CD4 positive cells and CD8 positive cells,31 the exact role of HSP60 remains unclear; however, it seems that the high immunogenic power of this protein is involved in the response.32 Zlotta et al. showed that repeated BCG bladder instillations induce increases in immunoglobulin G (IgG) antibody levels against tuberculin and also against several mycobacterial HSPs. These HSPs elicit cytotoxic T lymphocytes to bind HSP and HSP-associated peptides. In contrast, HSPs that are isolated from normal tissue do not possess this function. When these data were analyzed for correlations with clinical outcome, the results showed that increases in anti-HSP60 IgG levels after BCG therapy may be associated with a higher rate of bladder tumor recurrence.33

In the current study, HSP70 had the lowest expression levels (mean tumor expression, 44.8%), and no correlation was found with tumor grade, disease stage, and patient outcome. It has been suggested that HSP70 influences protein functions in bladder tumorigenesis, and it has been detected in association with the stabilization of mutant p53 protein.21 The increased stability of mutant p53 that is conferred by HSP may lead to a proliferative effect.34 Nevertheless, in patients with oral carcinoma, the overexpression of p53 is associated with a poor prognosis; however, the accumulation of p53 and HSP70 permits the formation of p53-HSP70 complexes, which reportedly are an essential prerequisite for eliciting a humoral immune response against p53 protein.35 In patients with bladder carcinoma, it is known that p53 overexpression is a marker of tumor aggressiveness36 and is correlated with high levels of metastasis. p53 is considered a prognostic marker; therefore, it is not surprising that the mean HSP70 expression in cystectomy specimens was only 10%. In 24 bladder carcinoma specimens from patients who underwent cystectomy exclusively for infiltrating tumors, Storm et al. found that 12 of 24 tumors (50%) had positive HSP27 expression and that 50% exhibited a loss of expression.37 In that specific population, the authors did not find any correlation between HSP27 expression with histologic grade, T stage, recurrence, metastases, or survival. Table 3 shows that in the current study, the mean HSP27 expression level in invasive tumors was 6.7%, and we observed a strong correlation with disease stage P < 0.001). The role of HSP27 in bladder carcinoma remains unknown, although it is known that the overexpression of HSP27 has strong prognostic value (correlated with poor survival) in patients with osteosarcoma, hepatocellular carcinoma, and esophageal squamous cell carcinoma.38–40 However, in patients with prostate carcinoma, Cornford et al. found a high correlation between the level of HSP27 expression and the Gleason score. Surprisingly, early in situ neoplastic transformation was associated with loss of HSP27 expression, whereas reexpression of this protein heralded a poor clinical prognosis.41

The current study demonstrates an important correlation between the loss of HSP60 and HSP90 expression and poor prognosis, and it suggests a potential role for these two proteins in the immune response. Our findings indicate that variations in HSP expression may lead to new, original, therapeutic approaches in the treatment of patients with bladder carcinoma. Constitutive and induced levels of HSP in three human bladder carcinoma cell lines were measured by Richards et al.,42 who demonstrated that those cells were capable of producing an induced heat shock response and increased HSP expression. These data indicate that sensitivity to immune response potentially may be induced by changes in HSP expression. Because HSP stimulates an immune response, overexpression may potentiate the host immune response; furthermore, HSP may be used as an antigen and may potentiate BCG immunotherapy.43, 44 Because HSPs are detected on the surface of TCC bladder tumors and function as accessory molecules in antigen presentation, it is possible that new therapies may be developed based on a vaccination against HSP.45, 46 Udono and Srivastava have shown that vaccination of mice with HSP-peptide complexes may induce immune responses against the peptide bound to HSP, but not to HSP itself.47 The presence of surface-expressed HSP and the development of HSP specific peptide immune complexes may provide unique opportunities to target the immune system in a variety of disease states. Influencing the activity of relevant HSPs in suitable patients may have therapeutic benefits.