Analysis of RNA-binding protein IMP3 to predict metastasis and prognosis of renal-cell carcinoma: a retrospective study

Articles

Analysis of RNA-binding protein IMP3 to predict metastasis and prognosis of renal-cell carcinoma: a retrospective study Zhong Jiang, Peigou G Chu, Bruce A Woda, Kenneth L Rock, Qin Liu, Chung-Cheng Hsieh, Cuizhen Li, Wengang Chen, Hai Ou Duan, Scott McDougal, Chin-Lee Wu

Summary Lancet Oncol 2006; 7: 556–64 Published Online June 14, 2006 DOI:10.1016/S1470-2045(06) 70732-X University of Massachusetts Medical Center, Department of Pathology and Department of Cancer Biology, Worcester, MA, USA (Z Jiang MD, Prof B A Woda MD, Prof K L Rock MD, Q Liu PhD, Prof C-C Hsieh ScD, C Li MD); City of Hope National Medical Center, Division of Pathology, Los Angeles, CA, USA (P G Chu MD, W Chen MS); and Massachusetts General Hospital, Department of Pathology and Department of Urology, Harvard Medical School, Boston, MA, USA (H O Duan MD, Prof S McDougal MD, C-L Wu MD) Correspondence to: Dr Zhong Jiang, Department of Pathology, Three Biotech, One Innovation Drive, Worcester, MA 01605, USA [email protected]

Background Distant metastasis is the main cause of death from renal-cell carcinoma, and the metastatic potential of tumours is often unpredictable. We aimed to investigate whether IMP3, an oncofetal RNA-binding protein, can be used as a biomarker to predict metastasis and prognosis of renal-cell carcinoma. Methods We studied 501 primary and metastatic renal-cell tumours. 371 patients with localised primary tumours were further investigated by use of survival analysis. We assessed IMP3 expression in tumour tissues by immunohistochemistry, and IMP3 mRNA and protein expression in selected tissues by quantitative real-time PCR and western blot analysis. Findings Compared with non-metastatic renal-cell tumours, IMP3 expression was greatly increased not only in metastatic tumours but also in a subset of primary tumours that were likely to subsequently develop metastases. Patients with primary localised tumours that did not express IMP3 had a longer metastasis-free survival and overall survival than did those with tumours expressing IMP3 (p<0·0001). Patients with IMP3-positive localised tumours had a much lower 5-year metastasis-free survival than did those with IMP3-negative tumours (for stage I tumours, 44% vs 98%, hazard ratio 17·18 [95% CI 7·82–37·78]; stage II, 41% vs 94%, 10·14 [3·46–29·68]; stage III, 16% vs 62%, 4·04 [2·23–7·31]). IMP3 expression was also associated with reduced 5-year overall survival (stage I, 32% vs 89%, 6·44 [3·63–11·42]; stage II, 41% vs 88%, 6·93 [2·63–18·27]; stage III, 14% vs 58%, 3·46 [1·98–6·05]). Multivariable analysis of IMP3 status (positive vs negative) in primary tumours showed hazard ratios of 5·84 (95% CI 3·60–9·49) for metastasis-free survival and 4·01 (2·66–6·05) for overall survival (both p<0·0001), which were much higher than hazard ratios associated with other independent risk factors. Interpretation IMP3 is an independent prognostic marker that can be used at initial diagnosis of renal-cell carcinoma to identify patients who have a high potential to develop metastasis and who might benefit from early systemic treatment.

Introduction Renal-cell carcinoma is the most common type of kidney cancer and accounts for about 85% of malignant kidney tumours.1,2 Incidence of the disorder has been rising steadily.3 About 36 160 new cases of kidney cancer were diagnosed in the USA in 2005, and about 12 660 patients will die from this disease.4 Resection of primary renal-cell carcinoma can be curative when the disease is localised. However, distant metastasis remains the main cause of treatment failure

and death from cancer.1,2 Patients with metastatic disease are typically given systemic treatment, which is associated with substantial toxic effects.1,2 Therefore, unless patients present with metastatic disease, clinical observation is the standard of care after nephrectomy. Currently, prediction of prognosis and selection of patients for postoperative adjuvant treatment is done mainly by pathological and clinical staging.5–8 However, because of substantial differences in the biological behaviour of renal-cell tumours classified in the same

Figure 1: IMP3 expression in primary and metastatic renal-cell carcinoma Immunohistochemical staining shows presence of IMP3 for bone metastasis (A) and primary tumour with subsequent development of metastasis (B), and absence of IMP3 in primary tumour without metastasis (C).

556

http://oncology.thelancet.com Vol 7 July 2006

Articles

IMP3 forward primer 5’-GCTAAAGTGAGGATGGTGATTATCACT-3’ IMP3 reverse primer 5’-ACTAACAAAGTTTTCTTCTTTAATTTTTCCAT-3’ IMP3 probe 5’FAM-ACCAGAGGCTCAGTTCAAGGCTCAGGGAA-TAMRA3’ GAPDH forward primer 5’-GAAGGTGAAGGTCGGAGTC-3’ GAPDH reverse primer 5’-GAAGATGGTGATGGGATTTC-3’ GAPDH probe 5’FAM-CAAGCTTCCCGTTCTCAGCC-TAMRA3’ FAM=carboxyfluorescein. TAMRA=carboxytetramethylrhodamine.

stage, the metastatic potential of localised tumours is difficult to predict. About 20% of patients with localised tumours develop metastasis, and the median survival for those with metastatic disease is roughly 13 months.2,9,10 Therefore, biomarkers that can accurately distinguish localised tumours with a high probability of metastasis from those that will remain indolent are needed. With such biomarkers, physicians can predict the patient’s prognosis and effectively target the individuals who would benefit most from adjuvant treatment. Molecular biomarkers have become a notable area of research in the study of renal-cell carcinoma. Various protein markers and gene-expression profiles based on DNA microarray analysis have shown potential in predicting disease outcome in patients with renal-cell carcinoma.11,12 IMP3 is a member of the insulin-like growth factor (IGF-II) mRNA binding protein (IMP) family that consists of IMP1, IMP2, and IMP3.13 IMP family members are important in RNA trafficking and stabilisation, cell growth, and cell migration during the early stages of embryogenesis.14 The IMP3 gene is located on chromosome 7p11.2 (within 11 cM from this location)15 and is identical to the KOC (KH-domaincontaining protein overexpressed in cancer) protein that was originally cloned from a pancreatic tumour cDNA screen.16 IMP3 is expressed in developing epithelium, muscle, and placenta during early stages of human and mouse embryogenesis, but is expressed at low or undetectable concentrations in adult tissues.13,14 IMP3 is also expressed in malignant tumours including soft-tissue sarcoma and cancer of the pancreas, lung, stomach, and colon, but is not detected in adjacent benign tissues.13,16–18 Moreover, IMP3 has been shown to promote proliferation of human leukaemia cells,19 suggesting that it is an oncofetal protein that could have a crucial role in regulating cell proliferation. However, http://oncology.thelancet.com Vol 7 July 2006

IMP3 expression in renal-cell carcinomas and the relation between IMP3 and tumour metastasis are unknown. In this study, we aimed to investigate whether IMP3 can be used as an independent biomarker to predict metastasis and prognosis in patients with renalcell carcinoma.

Methods Patients and tumour samples We obtained formalin-fixed, paraffin-embedded samples from 406 patients with primary renal-cell carcinoma who underwent radical or partial nephrectomy, from the archives at the University of Massachusetts Medical Center, Worcester, MA, USA (UMMC, n=159); Massachusetts General Hospital, Boston, MA, USA (MGH, n=152); and City of Hope National Medical Center, Los Angeles, CA, USA (CHNMC, n=95). These sources contained data for all patients for whom archival tissues and adequate clinical follow-up information were readily available. All cases were obtained between January, 1989, and December, 2003, and the diagnoses were confirmed by at least two pathologists. Staging was based on pathological findings according to the American Joint Committee on Cancer.20 By use of the tumour(T) node(N) metastasis(M) classification system, we identified 216 tumours at stage I (pT1a, pT1b), 64 at stage II (pT2), 98 at stage III (pT3a, n=62; pT3b, n=29; pT3, N1, n=7), and 28 at stage IV (pT2, N2 or M1, n=5; pT3, N2 or M1, n=12; pT4, N2 or M1, n=7; pT4, n=4). Follow-up for this retrospective study was undertaken by researchers (ZJ, PGC, and CLW) reviewing the clinical records of patients. We identified metastasis in samples from 119 (29%) of 406 patients with primary renal-cell tumours (UMMC, n=43 [27%]; MGH, n=44 [29%]; CHNMC,

100

80

Negative Positive

Proportion of patients (%)

Panel: Primers and probes used in quantitative rtPCR

60

40

20

0 Primary RCC without metastasis

Primary RCC with metastasis

Metastases from metastatic RCC

Type of tumour

Figure 2: IMP3 expression in patients with primary and metastatic renal-cell carcinomas (RCC)

557

Articles

Immunohistochemical analysis Western blot 1

2

3

4

5

6

70 kD IMP3 60 kD

40 kD

RCC with metastasis

RCC without metastasis

Metastatic RCC cell line

Actin

Quantitative western blot and rtPCR results

Ratio (%)

100

IMP3 to actin (western blot) IMP3 to GAPDH (rtPCR)

10

Quantitative analysis of immunostaining

1 1

2 IHC-positive

3

4

5 IHC-negative

Figure 3: Western blotting and quantitative rtPCR of IMP3 expression in renal-cell carcinoma (RCC) IHC=immunohistochemical analysis. Metastatic cell line of renal-cell carcinoma (lane 6) used for comparison. Actin used as loading control. Immunohistochemistry samples correspond with those in western blot. Ratios are of IMP3 protein to actin and IMP3 mRNA to GAPDH.

n=32 [34%]), during nephrectomy (n=30) or after surgery (n=89). 159 (39%) of 406 patients with primary renal-cell carcinoma died (UMMC, n=66 [42%]; MGH, n=53 [35%]; CHNMC, n=40 [42%]). An additional 95 samples of metastatic disease were also obtained for immunohistochemical analyses and compared with primary tumours, of which 26 were from the same patients with primary tumours and 69 from the biopsy or resection of metastatic tumours only. From all three institutions, these metastatic tumours were obtained from the lung (n=22); lymph nodes (n=10); gastrointestinal organs including liver, intestine, pancreas, and gallbladder (n=7); bone (n=23); brain (n=8); adrenal gland, thyroid, and ovary (n=10); head and neck (n=2); soft tissues (n=6); diaphragm, pleura, retroperitoneum, and omentum (n=7). The institutional review board at every institution approved this study. 558

We analysed 5-µm sections of formalin-fixed, paraffinembedded tissue from nephrectomy samples by using an avidin-biotinylated peroxidase complex as described elsewhere18 on the DAKO Autostainer (DAKO Corporation, Carpinteria, CA, USA). Sections of pancreatic carcinoma known to express IMP3 were used as positive controls for the L523S mouse monoclonal antibody, specific for IMP3 staining (Corixa Corporation, Seattle, WA, USA). For negative controls, we replaced the primary antibody with non-immune IgG. We defined positive staining of IMP3 as dark brown cytoplasm (figure 1A,B), and negative staining as no staining (figure 1C). IMP3 status was assessed by a genitourinary pathologist (ZJ) who was unaware of the clinical and pathological features of the cases or the clinical outcome. To assess the reproducibility of the immunohistochemical test for IMP3 expression, three other pathologists (CLW, PGC, and CL) independently assessed 50 cases that had been chosen randomly; all three pathologists fully concurred with the first assessment. Every positive sample was assessed further for the proportion of the cells that stained positively, and was scored as focal (≤30%) or diffuse (>30%).

To assess immunohistochemistry quantitatively, a pathologist (ZJ) analysed 270 different areas (15 different areas per sample) from nine cases showing IMP3 expression (ie, IMP3-positive) and nine cases without IMP3 expression (ie, IMP3-negative) by using a computerised image analyser (Automated Cellular Imaging System [ACIS], ChromaVision Medical System, San Juan Capistrano, CA, USA). Positive staining is calculated by the use of two thresholds; one that recognises blue background cells (ie, stained with haematoxylin) and another that recognises brown cells (ie, positive for IMP3 expression). The integrated optical density (IOD) is the total number of brown pixels multiplied by the brown intensity of those pixels. ACIS values were calculated as: IOD divided by the sum of the blue area and the brown area.

Western blotting analysis of IMP3 expression Frozen tissues of primary renal-cell carcinoma and a metastatic renal-cell-carcinoma cell line (ATCC number HTB-46, ATCC Global Bioresource Center, Manassas, VA, USA) were homogenised in three volumes of lysis buffer. Immunodetection was done with IMP3 monoclonal antibodies (L523S) at a 1 mg/L dilution with an enhanced chemiluminescence system (PerkinElmer Life Sciences, Boston, MA, USA). The membrane was stripped and reblotted with actin polyclonal antibodies (A-2066; Sigma, St Louis, MO, USA). Intensity of the signal was measured by densitometry software (National Institutes of Health http://oncology.thelancet.com Vol 7 July 2006

Articles

IMP3-positive tumours (n=71)

IMP3-negative p tumours (n=335)

Sex Female

17 (11%)

132 (89%)

Male

54 (21%)

203 (79%)

0·015

Age (years) Mean (SD)

61·8 (11·2)

59·5 (14·1)

0·28

Tumour stage I

22 (10%)

194 (90%)

II

11 (17%)

53 (83%)

III

24 (24%)

74 (76%)

IV

14 (50%)

14 (50%)

<0·0001

Statistical analysis

Tumour size (cm) Mean (SD)

8·5 (4·9)

6·1 (3·7)

<0·0001

30 (100%)

<0·0001

Tumour grade 1

0

2

17 (9%)

169 (91%)

3

36 (24%)

111 (76%)

4

18 (42%)

25 (58%)

Histological types Clear-cell type

58 (23%)

256 (74%)

Papillary

7 (11%)

56 (89%)

Chromophobe

4 (15%)

23 (85%)

Unclassified

2 (100%)

GAPDH—a housekeeping gene—was used as an internal reference. Frozen tissues were cut into 5-µm sections, and total RNA was extracted by Qiagen RNeasy Mini Kit (Qiagen, Valencia, CA, USA). We did the two-step quantitative rtPCR by using the ABI TaqMan PCR reagent kit (Applied Biosystems [ABI], Foster City, CA, USA) and IMP3 and GAPDH primers and probes for both genes on the ABI Prism 7700 system (panel). Expression of IMP3 mRNA was normalised with GAPDH mRNA expression, measured in the same RNA extraction and calculated as the ratio of IMP3 to GAPDH.

0·04

0

Percentages calculated by rows.

Table 1: Clinicopathological characteristics of patients with renal-cell carcinoma

[NIH] Image 1.61), and relative expression of IMP3 was normalised by the amount of the actin in every lane (calculated as the ratio of IMP3 to actin).

Quantitative real-time PCR (rtPCR) We measured mRNA amounts of IMP3 and GAPDH in renal-cell tumours by using quantitative rtPCR.

Overall survival was measured from the date of nephrectomy to the date of death, and was censored from the date of last follow-up for survivors. Metastasisfree survival was measured from the date of surgery to the date of first clinical evidence of metastasis, and was censored at the date of death or last follow-up for survivors. Data for age; sex; tumour size, stage, grade, and histological type; and IMP3 status were obtained as baseline variables. The distribution of every baseline variable was compared for IMP3-positive and IMP3-negative subgroups, with the Wilcoxon rank sum test for continuous variables and the Fisher’s exact test for categorical variables. Median follow-up was 63 months (range 1–174 months). 31 of 406 patients with metastatic disease identified during nephrectomy (stage III, n=7; stage IV, n=24) were excluded from the overall and metastasis-free survival analysis, because the aim was to assess the risk of metastasis after surgery in patients who presented with localised disease initially. By use of AJCC TNM staging criteria, we identified metastasis in 24 of 28 stage IV patients (M1, n=13; N2, n=10; N2 and M1, n=1) during nephrectomy. We excluded the remaining four patients with stage IV disease from the prognostic

A

B 1·00

1·00 IMP3-negative tumour 0·75 Overall survival

Metastasis-free survival

IMP3-negative tumour 0·75 p<0·0001 0·50 IMP3-positive tumour 0·25

0

0·50

p<0·0001

0·25

IMP3-positive tumour

0 0

1

2

3

4

5

6

7

8

9

0

10 11 12 13 14

Time after surgery (years) Number at risk IMP3-negative tumour IMP3-positive tumour

317 306 269 244 231 214 155 131 99 71 48 32 24 54 45 35 30 19 15 11 9 8 6 5 4 2

1

2

3

4

5

6

7

8

9

10 11 12 13 14

Time after surgery (years) 8

3

Number at risk IMP3-negative tumour IMP3-positive tumour

317 306 271 246 233 218 157 134 103 75 51 35 24 54 45 35 30 19 15 11 9 8 6 5 4 2

8

3

Figure 4: Kaplan-Meier analysis of metastasis-free survival (A) and overall survival (B) in patients according to IMP3 status

http://oncology.thelancet.com Vol 7 July 2006

559

Articles

A

Stage I

B

Stage I 1·00

IMP3–negative tumour

IMP3-negative tumour

0·75

0·75 Overall survival

Metastasis-free survival

1·00

p<0·0001 0·50 IMP3-positive tumour

IMP3-positive tumour

0·25

0·25

0

0

Number at risk IMP3-negative tumour IMP3-positive tumour

194 191 172 158 152 142 101 87 65 47 31 23 18 22 20 18 16

9

8

7

6

5

4

4

3

6

2

3

Number at risk IMP3-negative tumour IMP3-positive tumour

Stage II

C

p<0·0001

0·50

194 191 172 158 152 142 101 87 66 48 32 24 18 22

20 18 16

9

8

7

3

6

2

3

2

3

0·75 Overall survival

Metastasis-free survival

4

IMP3-negative tumour

0·75 p<0·0001 0·50 IMP3-positive tumour

p<0·0001 0·50 IMP3-positive tumour

0·25

0·25

0

0

53 53 47 45 41 39 28 21 17 16 12 11 10

E

9

8

6

4

3

2

8

6

3

2

2

Number at risk IMP3-negative tumour IMP3-positive tumour

53 53 48 46 42 40 29 23 19 18 14 9 11

F

Stage III

10

9

8

6

4

3

2

6

2

Stage III

1·00

1·00

0·75

0·75 IMP3–negative tumour

Overall survival

Metastasis-free survival

4

1·00 IMP3–negative tumour

0·50 p<0·0001 0·25

IMP3-negative tumour 0·50 p<0·0001 0·25

IMP3-positive tumour

IMP3-positive tumour

0

0 0

1

2

3

4

5

6

7

8

9

10 11 12 13 14

0

1

2

3

4

Time after surgery (years) Number at risk IMP3-negative tumour IMP3-positive tumour

5

Stage II

D

1·00

Number at risk IMP3-negative tumour IMP3-positive tumour

6

70

63 51 43 40 33 28 24 18 11

7

2

21

15 10

2

2

8

5

4

3

3

3

3

5

6

7

8

9

10 11 12 13 14

Time after surgery (years) 2

Number at risk IMP3-negative tumour IMP3-positive tumour

70

63 52 44 41 36 29 25 19 12

7

2

21

15 10

2

2

8

5

4

3

3

3

3

2

Figure 5: Kaplan-Meier analysis of metastasis-free and overall survival in patients with renal-cell tumours at stages I–III according to IMP3 status

analysis, since this small number would not be informative. Therefore, 371 patients with stage I–III disease and without metastasis during surgery were included in our survival analysis. 560

Overall survival and metastasis-free survival of 371 patients were estimated by the Kaplan-Meier method and assessed by the use of log-rank test for univariate analysis. We used the Cox proportional-hazard model to http://oncology.thelancet.com Vol 7 July 2006

Articles

Metastasis-free survival

Overall survival

Hazard ratio (95% CI)

Hazard ratio (95% CI)

p

p

IMP3 status (positive vs negative)

9·22 <0·0001 (6·01–14·14)

5·66 (3·93–8·16)

<0·0001

Age

1·01 (1·00–1·03)

0·167

1·03 (1·02–1·04)

<0·0001

Sex (female vs male)

0·77 (0·48–1·21)

0·250

0·94 (0·65–1·34)

0·722

Tumour size

1·16 (1·12–1·21)

<0·0001

1·10 (1·06–1·15)

<0·0001

II vs I

1·85 (0·96–3·53)

0·065

1·21 (0·72–2·02)

0·470

III vs I

6·05 (3·71–9·84)

<0·0001

3·39 (2·32–4·96)

<0·0001

2 vs 1

4·40 (0·60–32·4)

0·146

1·22 (0·52–2·87)

0·646

3 vs 1

11·37 (1·57–82·6)

0·016

2·59 (1·12–6·01)

0·027

4 vs 1

18·44 (2·44–139·6)

0·005

4·42 (1·78–10·96)

0·001

Papillary vs clear

0·48 (0·23–0·99)

0·047

0·66 (0·39–1·12)

0·121

Chromophobe vs clear

0·90 (0·39–2·08)

0·808

0·75 (0·35–1·61)

0·454

Tumour stage

Tumour grade

Histological type

was 44·3 in the IMP3-positive tumours and 0·01 in the IMP3-negative tumours (p=0·0025). We also used western blots to measure the reactivity of L523S antibody with IMP3 protein produced by renal-cell tumours (figure 3). L523S antibodies reacted with a protein at about 65 kD, the expected molecular weight for IMP3. IMP3 was highly expressed in the metastatic renal-cell-carcinoma cell line and in three primary tumours that had subsequent development of metastasis (these three primary tumours also showed positive results from immunohistochemical analysis). By contrast, no IMP3 expression was seen in two primary tumours, also with negative results from immunohistochemistry, which did not have subsequent development of metastasis (figure 3). We also detected overexpression of IMP3 mRNA in some primary renal-cell tumours. Figure 3 shows the results of quantitative rtPCR, corresponding to those of western blot analysis. The three primary tumours with subsequent development of metastasis had positive immunohistochemical analysis results, and showed overexpression of IMP3 mRNA compared with the two primary tumours that did not metastasise and had negative immunohistochemical analyses (figure 3). Results from the western blot and quantitative rtPCR of the same samples showed complete agreement between the tests (data not shown). However, since this analysis

Percentages calculated by rows.

Metastasis-free survival

Overall survival

Hazard ratio (95% CI)

p

Hazard ratio (95% CI)

p

IMP3 status (positive vs negative)

5·84 (3·60–9·49)

<0·0001

4·01 (2·66–6·05)

<0·0001

Age

1·00 (0·98–1·02)

0·752

1·02 (1·01–1·04)

0·006

Sex (female vs male)

1·08 (0·66–1·77)

0·747

1·08 (0·73–1·59)

0·694

Tumour size

1·11 (1·04–1·18)

0·001

1·05 (0·99–1·11)

0·081

II vs I

0·73 (0·30–1·77)

0·484

0·87 (0·44–1·71)

0·677

III vs I

3·15 (1·72–5·77)

<0·0001

2·08 (1·30–3·33)

0·002

2 vs 1

2·02 (0·27–15·37)

0·496

0·89 (0·37–2·16)

0·797

3 vs 1

4·43 (0·59–33·54)

0·149

1·57 (0·64–3·84)

0·319

4 vs 1

3·22 (0·40–25·80)

0·272

1·27 (0·47–3·44)

0·634

Papillary vs clear

0·50 (0·22–1·11)

0·090

0·71 (0·41–1·25)

0·236

Chromophobe vs clear

1·29 (0·54–3·08)

0·564

0·95 (0·44–2·09)

0·908

Table 2: Univariate analysis for metastasis-free and overall survival

assess the simultaneous contribution of baseline covariates in univariate and multivariable analyses. We regarded p<0·05 (two-sided) as significant.

Role of the funding source The sponsor of the study had no role in study design; data collection, data analysis, data interpretation; or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Tumour stage

Results

Tumour grade

We identified IMP3 protein in the cytoplasm (figure 1) of tumour cells. 59 (62%) of 95 metastases from renalcell tumours, 60 (50%) of 119 primary tumours with metastases during and after nephrectomy, and 11 (4%) of 287 primary tumours that were metastasis-free had IMP3 expression (figure 2). Of 71 IMP3-positive primary tumours, 38 had focal IMP3 positivity and 33 had diffuse positivity. No expression of IMP3 was seen in the benign kidney tissue adjacent to the tumours. Quantitative immunohistochemical analysis showed substantial differences in IMP3 staining values between positive and negative samples. The mean ACIS value http://oncology.thelancet.com Vol 7 July 2006

Histological type

Table 3: Multivariable analysis for metastasis-free and overall survival

561

Articles

was based on only five samples, a larger study is needed to show conclusive results. Table 1 provides the relevant clinical characteristics of the 406 patients with primary renal-cell carcinoma. Age was not associated with IMP3-positive status (p=0·28). Men were more likely than women to have tumours that expressed IMP3 (p=0·015). IMP3 expression was strongly associated with standard pathological predictors of clinical outcome (eg, tumour stage and grade, clear-cell histological type), and was identified mainly in large tumours. Only 10–24% of stage I, II, and III tumours expressed IMP3, whereas 50% of stage IV tumours expressed this protein. IMP3 was expressed mainly in high-grade (grade 3 and 4) tumours; all grade 1 tumours were negative for this protein. IMP3 positivity was slightly increased in conventional (clearcell) carcinomas (23%) compared with papillary (11%) and chromophobe (15%) types. Two unclassified tumours were both positive for IMP3. 371 patients who did not have metastasis at the time of surgery were included in follow-up. The proportion of samples with metastasis from primary renal-cell carcinoma after surgery differed substantially between patients who expressed IMP3 and those who did not. More patients with IMP3-positive tumours subsequently developed metastases than did those with IMP-negative tumours (IMP-positive, 43 [80%] of 54 patients, median follow-up 38 months [range 1–155]; IMP3-negative, 41 [13%] of 317, 70 [1–174]). Patients with IMP3-negative, primary renal-cell tumours had significantly longer metastasis-free survival and overall survival than those with IMP3positive tumours. 5-year metastasis-free survival was 89% (95% CI 85–92) in patients with IMP3-negative tumours versus 33% (20–46) in those with IMP3positive tumours (figure 4A). 5-year overall survival was 82% (77–86) in patients whose tumours did not express IMP3 versus 27% (16–40) in those whose tumours did express IMP3 (figure 4B). In patients with renal-cell tumours at stage I (figure 5A,B), II (figure 5C,D), and III (figure 5E,F), IMP3 expression was also associated with increased risk of metastasis and was strongly linked to poor overall survival. In the univariate analysis, the hazard ratio of IMP3 expression was 9·22 for metastasis-free survival and 5·66 for overall survival (table 2). Patients whose localised renal-cell tumours were IMP3-positive had a significantly lower 5-year metastasis-free survival than did those with tumours that were IMP3-negative for tumours of all stages (stage I, 44% vs 98%, hazard ratio 17·18 [95% CI 7·82–37·78]; stage II, 41% vs 94%, 10·14 [3·46–29·68]; stage III, 16% vs 62%, 4·04 [2·23–7·31]). 5-year overall survival was also significantly lower in patients with IMP3-positive tumours than in those with IMP3negative tumours (stage I, 32% vs 89%, 6·44 [3·63–11·42]; stage II, 41% vs 88%, 6·93 [2·63–18·27]; stage III, 14% vs 58%, 3·46 [1·98–6·05]). 562

The IMP3-positive staining patterns (focal vs diffuse) in the primary renal-cell tumours did not alter the patients’ prognosis. We saw no significant differences between patients with focal IMP3 staining and those with diffuse IMP3 staining in metastasis-free survival (p=0·732) or overall survival (p=0·728). No significant difference of overall survival was recorded between patients with IMP3-positive tumours and those with IMP3-negative tumours with stage IV disease (p=0·14). Table 3 shows the results of the multivariable analysis, which was stratified by three different centres in this study, for metastasis-free survival and overall survival in the 371 patients with localised disease at the time of surgery. For these analyses, all factors shown in table 1 were initially included in the model as potential risk factors. IMP3 expression in primary renal-cell tumours was a strong independent predictor of the patients’ clinical outcome; hazard ratios were 5·84 for metastasisfree survival and 4·01 for overall survival, which were much higher than those associated with all other independent risk factors (table 3). In addition to IMP3 status, age and tumour stage III (vs stage I) were significant risk factors for overall survival, and tumour size and stage III (vs stage I) were significant risk factors for metastasis-free survival (table 3).

Discussion We have shown that IMP3 is expressed much more in metastatic renal-cell tumours than in primary tumours, and that IMP3 expression in primary tumours can predict whether the tumour is likely to metastasise and can provide important prognostic information in patients with localised disease who undergo nephrectomy. Distant metastasis is the main cause of death and therapeutic failure in patients with renal-cell carcinoma.1,2 Although tumour stage, grade, and subtype provide some prognostic information, the metastatic potential of localised renal-cell carcinoma has often been unpredictable. Since a well-characterised antibody to IMP3, but not IMP1 or IMP2, was available for immunohistochemical study, we investigated IMP3 expression in primary and metastatic renal-cell tumours as part of our programme to develop biomarkers for clinical use. IMP3 displays several features that make it an attractive prognostic marker for renal-cell carcinoma. First, expression of IMP3 is associated with other known pathological indicators of aggressive renal-cell carcinoma. Our results showed that IMP3 expression was strongly associated with high tumour grade and stage, and with large tumour size. Second, IMP3 expression in primary renal-cell carcinoma is independently associated with poor clinical outcome. Patients with tumours that expressed IMP3 were four times more likely to die than those with tumours that did not express IMP3. Notably, the http://oncology.thelancet.com Vol 7 July 2006

Articles

multivariable Cox analysis showed that the hazard ratio for death in patients with primary tumours expressing IMP3 was much higher than those associated with any other clinical and pathological predictors, including: age; sex; and tumour stage, size, grade, and histological subtype. Third, IMP3 expression is an independent predictor of tumour metastasis. A reduced overall survival was strongly associated with metastasis in patients with IMP3-positive tumours. Our data showed that IMP3 expression was significantly increased not only in metastatic renal-cell tumours, but also in patients with primary tumours who developed metastatic disease, compared with renal tumours that did not develop metastasis. In patients with stage III disease, almost all patients with IMP3-positive tumours developed metastases after nephrectomy. In the multivariable Cox analysis, patients with IMP3-positive primary tumours were almost six times more likely to subsequently develop metastasis than were those with IMP3-negative tumours, after adjustment for other well-known clinical variables, such as tumour stage, size, grade, and subtype. Fourth, IMP3 immunohistochemical staining is a simple, inexpensive, and reliable assay. Since localised renal-cell tumours are usually treated by partial or radical nephrectomy, tumour tissue is routinely available for immunohistochemical staining with the monoclonal L523S antibody. Our findings showed that pathologists can readily analyse IMP3 immunohistochemistry without variation between observers to assess positive and negative staining, which can be easily used in routine clinical practice in all patients who have had nephrectomy. A computerised image analyser for quantitative immunohistochemistry (ACIS), which has been used clinically to investigate expression of ERBB2, also confirmed the accuracy of the assessment of IMP3 immunostaining by pathologists. In this study, we focused on assessment of IMP3 immunostaining by pathologists and used ACIS as a confirmatory test. A further study is needed to investigate how well quantitative image analysis of immunostaining results correlates with clinical outcome. Fifth, since patients with IMP3-expressing tumours have a high potential to develop metastasis, IMP3 provides a marker that can identify those who might benefit from a different follow-up approach after nephrectomy, and be used at initial diagnosis—the best time for considering early systemic treatment. Although IMP3 expression has been detected in many malignant tumours but not in adjacent benign tissues,11,12,15,16 and IMP3 could be crucial in the regulation of cell proliferation,17 but we still do not know much about the biological function of IMP3 in tumour pathogenesis. Our findings raise the possibility that IMP3, as an oncofetal protein, could have a direct role in the metastasis or more lethal behaviour of renal-cell http://oncology.thelancet.com Vol 7 July 2006

carcinoma. Notably, Yaniv and colleagues21 found that IMP3 expression in Xenopus laevis13 is needed for the migration of cells forming the roof plate of the neural tube and subsequently for neural-crest migration. In summary, IMP3 is an independent prognostic marker for renal-cell carcinoma. Expression of this protein in primary renal tumours could identify patients with early-stage disease, who have a high potential to develop metastasis after surgery and die from the disease. These findings could have therapeutic implications in patients who might benefit from early systemic treatment after nephrectomy.

If you would like to respond to an article published in The Lancet Oncology, please submit your correspondence online at: http://ees.elsevier. com/thelancetoncology

Conflicts of interest The University of Massachusetts Medical School has filed patent applications on the subject matter of this Article. The authors declare no other conflicts of interest. Contributors Z Jiang participated in the study design; data collection, analysis, and interpretation; writing of the report; and decision to submit for publication. P G Chu and C Li participated in the data collection, analysis and interpretation. B A Woda, K L Rock, and S McDougal participated in the data analysis and the writing of the report. Q Liu and C-C Hsieh undertook the statistical analysis. W Chen and H Duan participated in the data collection and analysis. C-L Wu participated in the data collection, analysis, and interpretation, and the writing of the report. Acknowledgments We thank Karen Dresser (University of Massachusetts Medical Center) for her technical support, Susan O’Hara (University of Massachusetts Medical Center) and Wenlei He (Massachusetts General Hospital) for clinical data collection, and Stephen Baker (University of Massachusetts Medical Center) for suggestions about statistical analysis. References 1 Motzer RJ, Bander NH, Nanus DM. Renal-cell carcinoma. N Engl J Med 1996; 335: 865–75. 2 Cohen HT, McGovern FJ. Renal-cell carcinoma. N Engl J Med 2005; 353: 2477–90. 3 Chow WH, Devesa SS, Warren JL, Fraumeni JF Jr. Rising incidence of renal cell cancer in the United States. JAMA 1999; 281: 1628–31. 4 Jemal A, Murray T, Ward E, et al. Cancer statistics, 2005. CA Cancer J Clin 2005; 55: 10–30. 5 Tsui KH, Shvarts O, Smith RB, et al. Prognostic indicators for renal cell carcinoma: a multivariate analysis of 643 patients using the revised 1997 TNM staging criteria. J Urol 2000; 163: 1090–95. 6 Sene AP, Hunt L, McMahon RF, Carroll RN. Renal carcinoma in patients undergoing nephrectomy: analysis of survival and prognostic factors. Br J Urol 1992; 70: 125–34. 7 Couillard DR, deVere White RW. Surgery of renal cell carcinoma. Urol Clin North Am 1993; 20: 263–75. 8 Thrasher JB, Paulson DF. Prognostic factors in renal cancer. Urol Clin North Am 1993; 20: 247–62. 9 Rabinovitch RA, Zelefsky MJ, Gaynor JJ, Fuks Z. Patterns of failure following surgical resection of renal cell carcinoma: implications for adjuvant local and systemic therapy. J Clin Oncol 1994; 12: 206–12. 10 Sandock DS, Seftel AD, Resnick MI. A new protocol for the followup of renal cell carcinoma based on pathological stage. J Urol 1995; 154: 28–31. 11 Lam JS, Shvarts O, Leppert JT, et al. Renal cell carcinoma 2005: new frontiers in staging, prognostication and targeted molecular therapy. J Urol 2005; 173: 1853–62. 12 Zhao H, Ljungber B, Grankvist K, et al. Gene expression profiling predicts survival in conventional renal cell carcinoma. PLoS Med 2006; 3: e13.

563

Articles

13

14

15

16

17

564

Nielsen J, Christiansen J, Lykke-Andersen J, et al. A family of insulin-like growth factor II mRNA-binding proteins represses translation in late development. Mol Cell Biol 1999; 19: 1262–70. Mueller-Pillasch F, Pohl B, Wilda M, et al. Expression of the highly conserved RNA binding protein KOC in embryogenesis. Mech Dev 1999; 88: 95–99. Monk D, Bentley L, Beechey C, et al. Characterisation of the growth regulating gene IMP3, a candidate for Silver-Russell syndrome. J Med Genet 2002; 39: 575–81. Mueller-Pillasch F, Lacher U, Wallrapp C, et al. Cloning of a gene highly overexpressed in cancer coding for a novel KH-domain containing protein. Oncogene 1997; 14: 2729–33. Wang T, Fan L, Watanabe Y, et al. L523S, an RNA-binding protein as a potential therapeutic target for lung cancer. Br J Cancer 2003; 88: 887–94.

18

19

20 21

Yantiss RK, Woda BA, Fanger GR, et al. KOC (K homology domain containing protein overexpressed in cancer): a novel molecular marker that distinguishes between benign and malignant lesions of the pancreas. Am J Surg Pathol 2005; 29: 188–95. Liao B, Hu Y, Herrick DJ, Brewer G. The RNA-binding protein IMP-3 is a translational activator of insulin-like growth factor II leader-3 mRNA during proliferation of human K562 leukemia cells. J Biol Chem 2005; 280: 18517–24. American Joint Committee on Cancer (AJCC). AJCC cancer staging manual, 6th edn. New York: Springer-Verlag, 2002. Yaniv K, Fainsod A, Kalcheim C, Yisraeli JK. The RNA-binding protein Vg1 RBP is required for cell migration during early neural development. Development 2003; 130: 5649–61.

http://oncology.thelancet.com Vol 7 July 2006