SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Materials and methods
  4. Results
  5. Discussion
  6. Perspectives
  7. Acknowledgments
  8. References
  9. Supporting Information

In malignant melanoma (MM) there is an urgent need to identify new markers with predictive value superior to the traditional clinical and histological parameters. Angiogenesis and lymphangiogenesis have been recognized as critical processes in tumour growth and metastasis development, and numerous studies have evaluated the significance of these parameters in predicting the prognosis in solid tumours, including MM. We set out to determine whether angiogenesis, lymphangiogenesis and lymphatic invasion (LI) are valuable prognostic markers in MM. We systematically reviewed the available literature and subsequently performed a meta-analysis on the compiled data. To be eligible for the systematic review, a study had to provide the microvessel density (MVD), the lymphatic vessel density (LVD) or information about LI, assessed by immunohistochemistry on the primary site in patients with MM. To be evaluable for the meta-analysis, a study also had to provide information on clinical outcome. We approached selected studies with the Reporting recommendations for tumour marker (REMARK) criteria, verifying whether they had followed the recommendations. In total, nine angiogenesis, seven lymphangiogenesis and 10 LI studies were included in our meta-analysis, representing 419, 474 and 802 patients, respectively. Using meta-analysis, we showed that peritumoral LVD and the presence of LI have prognostic value for patients with MM. In contrast, MVD and intratumoral LVD did not have prognostic value in these patients. LVD and LI seem to have prognostic value for patients with MM.

Malignant melanoma (MM) is a neoplastic disease with increasing incidence and limited therapeutic options for patients with metastatic disease. Numerous clinical trials in patients with advanced metastatic MM indicate that this disease is highly resistant to conventional cytotoxic chemotherapy, and only recently have systemic therapies been introduced that impact survival.[1] The introduction of BRAF and mitogen-activated protein kinase kinase inhibitors has made a substantial change for the treatment of patients with BRAF-mutated MM.[2] Similarly, new immunotherapeutic approaches interfering with immune cell regulation, such as ipilimumab,[3] and agents interfering with the programmed death ligand 1 axis[4] seem to be very effective and promising in certain subgroups of patients. The most widely used prognostic indicators for survival in MM are the Breslow index, ulceration and the mitotic index. However, these are still inaccurate for a significant number of patients. Up to 15% of patients with invasive thin tumours (< 1 mm) still develop metastatic disease.[5] Similarly, a substantial proportion of patients with thick melanoma (≥ 4 mm) will have relatively long-term survival (58% for > 5 years).[6] Therefore there is an urgent need to identify new prognostic indicators and effective treatment options to improve outcomes for patients with MM.

One of the key biological processes that appears to be relevant to the establishment and progression of most forms of cancer, including MM, is angiogenesis, the generation of new capillary blood vessels from pre-existing vessels. Tumour angiogenesis was shown to be associated with poor patient outcome in a variety of cancers[7-10] including MM.[11-15] The measurement of this parameter is complicated by the fact that this is a dynamic process. Most studies have focused on the product of angiogenesis, microvessel density (MVD). Counting microvessels in a microscopic field gives an estimate of the net result of phases of angiogenesis and of blood vessel remodelling or regression at a particular time point. The reason for counting microvessels in vascular ‘hot spots’, or in areas of high vascular density within the tumour, identified by an inspection of the tumour section at low magnification (as defined by Weidner et al.),[16] is that these regions are thought to represent areas of biological importance. As the microvessels originate from tumour cell clones with the highest angiogenic potential, they provide a means of accessing the circulation, and hence increase the probability of vascularized metastasis. Hypoxia is considered a strong angiogenic drive.[17, 18] In addition, given the heterogeneity of tumour vascularization, obtaining a reproducible result in MVD quantification implies the evaluation of large tumour areas.[19] Welter et al.[20] formulated a theoretical model to analyse the vascular remodelling process of an arteriovenous vessel network during solid tumour growth. The authors explain the hot spot formation through blood pressure gradients that arise in the tumour vasculature when short cuts between nearby arteries and veins are formed.

It is of interest that MM of the skin is distinguished by its propensity for early metastatic spread via lymphatic vessels to regional lymph nodes, even at the early stages of tumour invasion. Based on a cohort of 431 patients with MM with a mean tumour thickness of 1.89 mm, Essner et al.[21] reported that 21% of patients presented with lymph-node metastases at the time of diagnosis. Lymph-node metastases, as detected by analysis of the sentinel lymph node, is one of the major determinants for the staging and clinical management of MM,[22] although disease will recur in 22% of all lymph-node-negative patients, and 15% will die within 5 years of initial treatment of the primary tumour.[23] Recent evidence suggests that tumour-associated lymphangiogenesis promotes MM metastasis to lymph nodes.[15, 24-27] The methodology of lymphangiogenesis quantification is based on the assumption that a functional increase in lymphatic vessels during tumour growth also occurs in hot spots.[28] As hot spots are considered to be a response to the local release of growth factors stimulated by localized changes in oxygen, and the data on the association of lymphangiogenesis with hypoxia are still contradictory, the relevance of counting lymphatic vessels in hot spots has been questioned.[29] However, using comparative in vitro analysis of single-cell clones derived from an MM of a single patient, Swoboda et al.[30] recently demonstrated that the lymphangiogenic potential is heterogeneously distributed among MM cells within one given tumour. This lymphangiogenic phenotype was found to be associated with high expression of vascular endothelial growth factor (VEGF)-C and platelet-derived growth factor-C,[30] which could explain the distribution of tumour lymphatic vessels in hot spots. Supporting these data, Dadras et al.[15, 27] found lymphatic vessels distributed in prominent hot spots both within and around primary cutaneous MM.

Despite the fact that many observational studies (either retrospective or prospective) have concluded that MVD or lymphatic vessel density (LVD) are prognostic factors in MM, other studies have failed to support these conclusions, probably due to the lack of consensus methodology. To determine whether angiogenesis, lymphangiogenesis and/or lymphatic invasion (LI) are prognostic indicators in human cutaneous MM, we undertook a systematic review of the literature with a meta-analysis. We report the results of all studies identified in the literature assessing MVD, LVD or LI in human cutaneous MM, and we performed the meta-analysis on those studies fulfilling the inclusion criteria (vide infra). We also analysed to what extent each publication meets the Reporting recommendations for tumour marker (REMARK) criteria. The REMARK checklist consists of 20 items that must be considered when publishing tumour marker prognostic studies. It was developed to encourage transparent and complete reporting so that the relevant information will be available to others to help them judge the usefulness of the data and to understand the context in which the conclusions apply.[31]

The aim of this meta-analysis was to test the hypothesis that MVD, LVD and LI would be able to predict the development of metastases in a global population of surgically treated patients with MM, and thus would facilitate the conversion of MVD, LVD and LI into accepted prognostic factors in patients with MM.

Materials and methods

  1. Top of page
  2. Summary
  3. Materials and methods
  4. Results
  5. Discussion
  6. Perspectives
  7. Acknowledgments
  8. References
  9. Supporting Information

Publication search and selection

In order to collect material for this meta-analysis, a computerized search of the PubMed database of the National Library of Medicine was performed to identify all relevant articles published up to 1 October 2012 using the following Medical Subject Heading terms: for angiogenesis in melanoma: ‘melanoma’ AND (‘neovascularisation, pathologic’ OR ‘microvessels’) and for lymphangiogenesis in melanoma: ‘melanoma’ AND (‘lymphangiogenesis’ OR ‘lymphatic vessels’). To be eligible for our meta-analysis, studies had to deal with cutaneous MM only, to assess MVD, LVD or LI by immunohistochemistry, to measure MVD, LVD or LI in the primary tumour and to evaluate the correlation between these parameters and the development of metastasis or survival. More detailed information is provided in file Data S1a (see Supporting Information).

Methodological assessment

For the meta-analysis the information was extracted from all publications using a standardized data collection method. We did not set a predefined minimum number of patients for a study to be included in our meta-analysis, or a minimum duration of follow-up. We did not attribute a particular weight to each study by a quality score, because no such score has received general approval, especially for observational studies.[32] When duplicated studies were identified, we included in our meta-analysis the study involving the highest number of patients from which data could be extracted (usually the latest). The decision to exclude a particular study was always taken prior to any data analysis. Studies included in this systematic review were termed ‘eligible’, and those providing sufficient data for the meta-analysis were termed ‘evaluable’. Detailed information is provided in file Data S1b (see Supporting Information).

Statistical methods

Statistical tests were carried out comparing mean MVD or LVD between the groups with metastatic and nonmetastatic melanomas, or comparing survival distributions between the groups with high and low MVD or LVD (usually with median MVD or LVD as cut-off). A study was considered significant with a P-value < 0.05 in univariate analysis. A study was termed ‘conclusive’ when the microvessel count predicted the development of metastasis or poorer survival, and ‘nonconclusive’ if no statistical difference between the two groups was detected. Not all studies reported results as microvessels per mm[2]. Taking into account that not all melanomas are of the same size and that there can be differences in the field size depending on the microscope used, we normalized the number of blood and lymphatic vessels to a measurement per mm[2]. The meta-analysis was performed using a random effects model using the mean difference for the measurements between the metastatic and the nonmetastatic groups (i.e. effect). In order to avoid drawing conclusions using a nonrepresentative population, we performed the comparison between those studies included in the meta-analysis and those we excluded due to insufficient data. We also determined whether there was an influence of different features of the studies (sample size, mean follow-up, the number of the observers, the methodology of vessel quantification and the antibody used) on its final result (positive vs. negative). All statistical calculations for our meta-analysis were performed in Bioconductor using R version 2.13.0 (Foundation for Statistical Computing, Vienna, Austria) (http://www.bioconductor.org/). More detailed information is provided in file Data S1c (see Supporting Information).

Reporting recommendations for tumour marker criteria

For each of the prognostic studies evaluating angiogenesis, lymphangiogenesis and LI in MM, we evaluated the REMARK criteria.[31] For each study, we calculated a score that indicates how well the REMARK criteria were followed,[33] as previously described.[28] A score of 1 (green colour) was given for each REMARK recommendation that was completely followed. A score of 0 (red colour) was given for each REMARK recommendation that was not considered in the study. Finally, a score of 0.5 (yellow colour) was given for each REMARK recommendation that was partially followed. Subsequently a general score on how well the guidelines were followed and the percentage of the maximum score were calculated for each study.

Results

  1. Top of page
  2. Summary
  3. Materials and methods
  4. Results
  5. Discussion
  6. Perspectives
  7. Acknowledgments
  8. References
  9. Supporting Information

Study selection and characteristics

The initial search yielded 1124 publications for angiogenesis and 92 publications for lymphatic vessels (lymphangiogenesis and LI). After screening the titles and abstracts for relevance and assessing full-text articles, 28 publications on angiogenesis (published between 1988 and 2012), 14 publications on lymphangiogenesis (published between 1997 and 2012) and 15 publications evaluating LI (published between 2005 and 2012) were selected. More details on the study selection are provided in file Data S1a (see Supporting Information). Finally, 22 studies evaluating MVD,[11-15, 27, 34-49] 12 studies evaluating LVD[5, 12, 13, 27, 29, 49-55] and 12 studies evaluating LI[5, 13, 27, 29, 49, 54, 56-61] were found to be eligible for inclusion in this analysis. The design was more often a retrospective study (20 angiogenesis, 11 lymphangiogenesis and 11 LI studies) than a prospective cohort study (two angiogenesis, one lymphangiogenesis and one LI study). The individual characteristics of the eligible studies for the systematic review are reported in Tables 1, 2 and 3.

Table 1. Main characteristics of the 22 angiogenesis studies
StudyStudy designNo. patientsMean or median age (years)Mean or median Breslow (mm)Mean or median follow-up (months)MVD assessmentAntibodyMagnification usedNo. hot spots examinedBlinded readingNo. observersResultSeparate count for intra- and peritumoral vessels?
  1. C, conclusive; Met, metastatic; MVD, microvessel density; NC, not conclusive; NM, nonmetastatic; Prosp, prospective; Retro, retrospective; ?, information not provided in the article. aStudy included in the final meta-analysis.

Srivastava 1988[34],Retro2056.0 Met; 54.0 NM2.18 Met; 2.07 NM111.0AutomatedUEA-1× 4003–18NCNo
Carnochan 1991[35]Retro107??60Chalkley countingUEA-1× 200???NCNo
Barnhill 1994[36]Retro12053.2 Met; 55.9 NM2.2 Met; 2.0 NM106.8Weidner's method and automatedUEA-1× 4002–3Yes?NCNo
Graham 1994[37],Retro37???Weidner's methodvWF, CD34× 250?Yes3NCNo
Busam 1995[38],Retro12053.2 Met; 55.9 NM2.2 Met; 2.0 NM106.8Weidner's method and automatedUEA-1× 400?Yes2NCNo
Guffey 1995[39]Retro1655.0 Met; 43.0 NM0.51 Met; 0.38 NM?Weidner's method and automatedCollagen IV× 400, × 2003Yes2NCOnly on tumour base
Marcoval 1996[40]Retro125??64.0Weidner's methodUEA-1× 400???NCNo
Ilmonen 1999[41]Retro84?1.248.6Weidner's methodCD31????COnly intratumoral
Neitzel 1999[42],Retro2567.4 Met; 63.0 NM??Weidner's methodvWF× 400?Yes2CNo
Straume 1999[11]Retro88??90.0Weidner's methodvWF× 40010?2CNo
Dawn 2002[43]Retro24???SemiquantitativeCD105??2NCNo
Kashani-Sabet 2002[44]Retro41747.02.48> 48.0Semiquantitative??1CNo
Massi 2002[45],Retro45?? Met; 4.3 NM94 Met; 154 NMAutomatedCD31× 100?NCNo
Dadras 2003[27]Retro3754.9 Met; 53.8 NM2.6 Met; 2.5 NM78.0AutomatedCD31× 1003NCNo
Ribatti 2003[46],Prosp25???Weidner's methodvWF× 250??2CNo
Valencak 2004[12]Retro12057.1?44.6Weidner's methodCD31??Yes2COnly intratumoral
Dadras 2005[13],Prosp4554.0 Met; 53.8 NM3.01 Met; 1.70 NM?AutomatedCD31× 1003CYes
Depasquale 2005[14]Retro204??> 48.0Chalkley countingCD31× 2003?1CYes
Hillen 2006[47]Retro5852.7?> 120Weidner's methodCD31, CD34, CD31/Ki67× 2004?2NCYes
Demirkessen 2006[15],Retro50?3.5 Met; 2.7 NM36Weidner's methodCD31× 400???CNo
Kiss 2007[48],Retro52??60Weidner's methodCD34× 2003??NCYes
Storr 2012[49],Retro20253.0?38.0Chalkley countingCD34?Yes2NCNo
Table 2. Main characteristics of the 12 lymphangiogenesis studies
StudyStudy designNo. patientsMean or median age (years)Mean or median Breslow (mm)Mean or median follow-up (months)LVD assessmentAntibodyMagnification usedNo. hot spots examinedBlinded readingNo. observersResultSeparate count for intra- and peritumoral vessels?
  1. C, conclusive; LVD, lymphatic vessel density; Met, metastatic; MSI, microsatellite instability; NC, not conclusive; NM, nonmetastatic; Prosp, prospective; Retro, retrospective; ?, information not provided in the article. aStudy included in the final meta-analysis.

de Waal 1997[50]Retro27???Weidner's methodCD31/PAL-E, CD34, Anti-COL-IV× 100??2NCNo
Dadras 2003[27],Retro3754.9 Met; 53.8 NM2.6 Met; 2.5 NM78.0AutomatedLYVE-1× 1003CYes
Straume 2003[51],Retro202??76.0Weidner's methodLYVE-1× 4005?1CYes
Shields 2004[29],Retro21?3.4 Met; 2.3 NM68.4 Met; 99.6 NMWeidner's methodLYVE-1× 400?Yes1CYes
Valencak 2004[12]Retro12057.1?44.6Weidner's methodD2-40??Yes2CNo
Dadras 2005[13],Prosp4554.0 Met; 53.8 NM3.01 Met; 1.70 NM?AutomatedLYVE-1, D2-40× 1003CYes
Sahni 2005[52]Retro36???Chalkley countLYVE-1× 4003Yes2NCYes
Massi 2006[53],Retro4563.3 Met; 55.3 NM3.0 Met; 2.9 NM42.0AutomatedD2-40× 2003CYes
Xu 2008[54]Retro106??> 120.0MSI analysisD2-40× 200?Yes2CYes
Emmett 2010[5],Retro10257.23 Met; 55.63 NM2.4 Met; 1.9 NM60.0Weidner's methodLYVE-1× 400?Yes3COnly peritumoral
Storr 2012[49]Retro20253.0?38.0Weidner's methodD2-40?Yes2NCNo
Shayan 2012[55],Retro2264.2 Met; 69.7 NM4.1 Met; 4.2 NM?AutomatedD2-40× 100?Yes2CYes
Table 3. Main characteristics of the 12 lymphatic invasion (LI) studies
StudyStudy designNo. patientsMean or median age (years)Mean or median Breslow (mm)Mean or median follow-up (months)Blinded readingNo. observersPatients with LI,%AntibodyResultInformation about intra- and peritumoral lymphatic invasion?
  1. C, conclusive; Met, metastatic; NC, not conclusive; NM, nonmetastatic; Prosp, prospective; Retro, retrospective; SLN, sentinel lymph node; ?, information not provided in the article. aStudy included in the final meta-analysis.

Dadras 2003[27],aRetro18 SLN+; 19 SLN−54.9 Met; 53.8 NM2.6 Met; 2.5 NM78.011 Met; 0 NMLYVE-1CNo
Shields 2004[29],aRetro13 Met+; 8 Met−?3.4 Met; 2.3 NM68.4 Met; 99.6 NMYes169 Met; 12 NMLYVE-1CYes
Dadras 2005[13],aProsp18 SLN+; 27 SLN−54.0 Met; 53.8 NM3.01 Met; 1.70 NM?11 Met; 0 NMLYVE-1, D2-40CNo
Niakosari 2008[56],aRetro23 SLN+; 73 SLN−???Yes265 Met; 23 NMD2-40CNo
Xu 2008[54]Retro36 Met+; 70 Met−??> 120.0Yes233D2-40CNo
Petitt 2009[57],aRetro5 SLN+; 22 SLN−58.03.3?Yes240 Met; 36 NMD2-40NCNo
Doeden 2009[58],aRetro23 SLN+; 34 SLN−54.02.535.2Yes239 Met; 9 NMLYVE-1, D2-40CNo
Petersson 2009[59]Retro25 SLN+; 11 SLN−?1.0757.6?155 Met; 12 NMD2-40CYes
Emmett 2010[5],aRetro10 SLN+; 8 SLN−57.23 Met; 55.63 NM2.4 Met; 1.9 NM60.0Yes380 Met; 38 NMLYVE-1COnly peritumoral
Fohn 2011[60],aRetro18 SLN+; 46 SLN−50.01.20?Yes167 Met; 4 NMD2-40CNo
Storr 2012[49],aRetro120 SLN+; 66 SLN−53.0?38.0Yes228 Met; 26 NMD2-40NCNo
Xu 2012[61],aRetro72 Met+; 179 Met−??> 120.0Yes265 Met; 34 NMD2-40CNo

Report of the study results

Among the 22 angiogenesis studies, nine studies (1058 patients) found that MVD was able to predict metastasis development in patients with MM (so-called conclusive studies), whereas 12 studies (926 patients) found no relation between MVD and clinical outcome (nonconclusive studies). In one study (37 patients) the MVD had prognostic value only in thin but not in thick MM tumours.[37] Among the 12 lymphangiogenesis studies, nine (700 patients) found that LVD could predict metastasis development in patients with MM (conclusive studies), whereas three studies (265 patients) found no relation (nonconclusive studies). Among the 12 studies focusing on lymphatic vessel invasion, 10 studies (731 patients) found that LI could predict metastasis development in patients with MM (conclusive studies), whereas two studies (213 patients) found no relation (nonconclusive studies). We evaluated the key features of all the studies we selected for the review and the influence of these characteristics on the final result of the study (positive or negative). More detailed information is provided in file Data S2b (see Supporting Information).

Meta-analysis

In order to avoid drawing conclusions using a nonrepresentative study population we compared the key features of those studies included in the meta-analysis with those we excluded from the meta-analytical calculations. Among 22 studies assessing angiogenesis in melanoma, nine independent studies fulfilled our inclusion criteria for meta-analysis, representing 419 patients. The excluded studies are listed in Table S1 (see Supporting Information), with information on the reasons for rejection. The reason to exclude the studies using Chalkley counting or semiquantitative methods was because these procedures give different data distributions from the Weidner method. The number of studies reporting Chalkley counts was too small to perform a separate meta-analysis. With a random effects model the total mean MVD difference between the metastatic and nonmetastatic groups was not statistically significant (> 0.05) (Fig. 1a). For more detailed information see file Data S2c (see Supporting Information).

image

Figure 1. Results of the meta-analysis of (a) nine studies evaluating microvessel density (MVD), (b) seven studies evaluating peritumoral lymphatic vessel density (LVD), (c) four studies evaluating intratumoral LVD and (d) 10 studies evaluating lymphatic invasion (LI). The size of the boxes is proportional to the weights of the study, determined by the number of patients included. The centre of the diamond gives the combined mean MVD, peritumoral LVD and intratumoral LVD difference, and the combined odds ratio for LI in the metastatic vs. nonmetastatic groups in (a–d), respectively.

Download figure to PowerPoint

Among the 12 studies evaluating lymphangiogenesis, seven independent studies assessing LVD fulfilled our inclusion criteria, representing 474 patients. Excluded studies are listed in Table S1, with information on the reason for rejection. All of the seven included studies provided a mean peritumoral LVD in the metastatic and nonmetastatic groups, but only four of these studies provided a mean intratumoral LVD. With a random effects model the total mean peritumoral LVD difference between two groups was statistically significant (< 0.05) (Fig. 1b). The mean intratumoral LVD difference between the two groups was not considered significant (> 0.05) (Fig. 1c). For more detailed information see file SF2c. Among the 12 studies evaluating LI, 10 independent studies fulfilled our inclusion criteria, representing 802 patients. Excluded studies are listed in Table S1. Using a random effects model for LI the total odds ratio difference of the logarithmic transformation between the two groups was statistically significant (< 0.05) (Fig. 1d).

Reporting recommendations for tumour marker criteria

The angiogenesis studies achieved between 18% and 84% of the maximum score. Scores were < 50% in nine of the 22 studies (Fig. 2). The lymphangiogenesis studies achieved between 33% and 84% of the maximum score, with two of the 12 studies obtaining < 50% of the maximum score. The LI studies achieved between 40% and 79% of the maximum score, with only one of the 12 studies obtaining < 50% of the maximum (Fig. 3).

image

Figure 2. List of studies on the prognostic value of angiogenesis in melanoma, showing how well the Reporting recommendations for tumour marker (REMARK) criteria were followed. The REMARK criteria did apply (green), did not apply (red) or only partly applied (yellow) to the study. CI, confidence interval; NA, not applicable.

Download figure to PowerPoint

image

Figure 3. List of studies on the prognostic value of lymphangiogenesis and lymphatic invasion in melanoma, showing how well the Reporting recommendations for tumour marker (REMARK) criteria were followed. The REMARK criteria did apply (green), did not apply (red) or only partly applied (yellow) to the study. CI, confidence interval; NA, not applicable.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Materials and methods
  4. Results
  5. Discussion
  6. Perspectives
  7. Acknowledgments
  8. References
  9. Supporting Information

Our overview and meta-analysis show that peritumoral LVD and the presence of LI predict the development of metastatic disease in MM of the skin. Our findings are in agreement with results published by Wang et al.,[24, 25] who reported an association between LVD and prognosis in breast and nonsmall-cell lung carcinomas. In our analysis MVD and intratumoral LVD did not have a predictive value for the development of metastatic disease in MM. These findings are supported by the results of a meta-analysis published by Trivella et al.[8] on the prognostic significance of MVD in nonsmall-cell lung carcinoma. In contrast, MVD was found to be a prognostic factor in nonsmall-cell carcinoma, breast cancer and colorectal cancer in the meta-analyses performed by Meert et al.,[9] Uzzan et al.[7] and Des Guetz et al.[10] To our knowledge, ours is the first meta-analysis on the significance of LI in the prediction of metastasis development in human cancer.

Our findings are in agreement with the fact that many human cancers, including MM, metastasize first via lymphatic vessels to the sentinel lymph node. The first specific growth factors discovered were VEGF-C and VEGF-D, which bind to the receptor VEGFR-3, expressed on lymphatic endothelium.[62] Using a xenotransplant transgenic tumour model in mice, Rinderknecht and Detmar demonstrated that tumour-induced lymphangiogenesis plays an important and active role in the promotion of cancer metastasis to lymph nodes.[62] Another very interesting observation made by that group was that stable overexpression of VEGF-C or VEGF-D by cancer cells potently enhanced tumour-associated lymphangiogenesis, leading to an increased incidence of lymph-node metastasis in MM.[63] VEGF-C expression levels were reported to be significantly correlated with LVD[13] and with lymph-node metastasis in primary MM.[13, 64-66] Moreover, primary MMs in the vertical growth phase were found to express more VEGF-C than those in the horizontal growth phase.[66] Surprisingly, the induction of lymph-node lymphangiogenesis by tumour cells was found significantly to promote distant metastasis.[67] Together with our results, these findings suggest the importance of lymphangiogenesis as a novel prognostic indicator for the risk of metastasis, and also as a promising therapeutic target in patients with MM.

We were not able to demonstrate the prognostic significance of angiogenesis in patients with MM. The lack of correlation between MVD and MM metastasis can be explained, at least in part, by the use of nonstandardized methodology of angiogenesis quantification. Although most studies used a technique similar to that of Weidner et al.,[16] many variations in MVD assessment exist. The computerized image analysis system is an automated counting technique that improves reproducibility and reduces interobserver variability, and has been proposed as a more objective method of assessing MVD.[18] However, this procedure is time consuming and expensive, making it difficult to use in routine pathological analyses of tumour samples. Chalkley counting is considered to be a simple, reproducible and acceptable procedure for daily clinical use and is recommended by international experts in tumour angiogenesis and lymphangiogenesis quantification. As no decisions have to be made on whether adjacent stained structures are stained microvessels or not, Chalkley point counting should be a more objective approach. The additional advantage of the Chalkley counting method is that it is not influenced by the field area examined, making this technique more reproducible. Obviously, the choice of endothelium marker for immunohistochemical staining might influence the conclusions. It is widely known that different markers have different sensitivity and specificity, which could influence the final result of the study.[17-19, 28, 68] Thus, the most appropriate solution would be to perform separate analyses for each antibody. However, this was not possible due to the small number of studies using each endothelium marker.

If additional studies confirm this finding, the lack of prognostic significance of angiogenesis could indicate the existence of angiogenesis-independent growth in MM tumours, explaining the frequent failure of anti-VEGF drugs in this tumour type. Therefore, the hypothesis about the existence of alternative, angiogenesis-independent vascularization mechanisms in MM, which would not be detected using endothelial markers, should be tested.

Our study has a number of limitations. Our results are based on an aggregation of data from mostly retrospective and observational studies, which are more prone to bias than prospective trials. Additionally, the present study collected data from published trials rather than collecting individual patient data. We were not able to perform quantitative aggregation of survival data because most studies were based on comparing mean MVD or LVD or the presence of LI between groups with metastatic vs. nonmetastatic MM, and did not provide information about the link of these parameters to overall and disease-free survival. A considerable majority of the studies drew conclusions based on small numbers of included patients. In those studies in which multivariate analyses were performed, the factors controlled for were few and differed between studies. If we look at the REMARK criteria colour maps of those studies reporting the prognostic significance of MVD, LVD or LI in MM included in the review, we can appreciate the predominance of the yellow and red colours, indicating that in many publications the reported information was insufficient.

Perspectives

  1. Top of page
  2. Summary
  3. Materials and methods
  4. Results
  5. Discussion
  6. Perspectives
  7. Acknowledgments
  8. References
  9. Supporting Information

Increasing evidence indicates that tumour-associated lymphangiogenesis promotes MM metastasis to lymph nodes, representing a novel prognostic indicator. Moreover, the recently discovered importance of lymph-node lymphangiogenesis, induced even before the tumour has metastasized, suggests that lymphangiogenesis also represents a novel target for early detection and treatment of MM metastasis.[62]

Additional studies on the prognostic value of angiogenesis in MM need to be performed, taking into account the angiogenesis-independent vascularization mechanisms. As the difference in signals regulating angiogenesis between the primary tumour and metastasis has been suggested as one of the mechanisms of resistance against VEGF (receptor) blockade,[69] a detailed analysis of the vascular network in MM metastasis should be performed.

Two international consensus studies on quantification of angiogenesis[17, 19] and one on the quantification of lymphangiogenesis[28] have been published. To obtain comparable parameters from different studies, the authors should follow the recommendations given by international experts and also take into account the REMARK criteria[31] concerning the publication of the results.

Acknowledgments

  1. Top of page
  2. Summary
  3. Materials and methods
  4. Results
  5. Discussion
  6. Perspectives
  7. Acknowledgments
  8. References
  9. Supporting Information

We thank all of the authors whose publications could be included in our meta-analysis, and especially those who provided us with complementary data allowing meta-analytical calculations.

References

  1. Top of page
  2. Summary
  3. Materials and methods
  4. Results
  5. Discussion
  6. Perspectives
  7. Acknowledgments
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Materials and methods
  4. Results
  5. Discussion
  6. Perspectives
  7. Acknowledgments
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
bjd12688-sup-0001-SupplementaryData.docxWord document187K

Data S1. Materials and methods. (a) Publication search and selection, (b) methodological assessment, (c) statistical methods.

Data S2. Results. (a) Study selection and characteristics, (b) study results report, (c) meta-analysis.

Table S1 Excluded studies

References for supplementary files.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.