Endothelial cell proliferation and vascular patterns in urothelial carcinoma

Abstract

Introduction: The bladder cancer has some characteristics: the sixth most incident neoplasm in the United States, the majority of diagnosed cases in those 55 years of age and older, four times more common in man than women, a reduced five-year survival rate in case of metastatic disease. Despite the beneficial effects of the combination therapy and immunotherapy, the low response rate and drug resistance were reported. The main goal of this work was evaluation of the endothelial cell proliferation from urothelial carcinomas. Patients, Materials and Methods: Fifty-two cases of T2–T4 infiltrative bladder tumors, aged between 46 and 78 years, were investigated. Morphological, simple and cluster of differentiation 31 (CD31)/Ki67, CD31/smooth muscle actin (SMA) double immunostaining were performed. Results: In all the analyzed infiltrative bladder tumors, three types of vessels were noticed: immature, intermediate and mature. In the central part of the tumor area, the following distribution of vessel types was noticed: immature (62.25%), intermediate (35.1%), and mature vessels (2.65%). In the peripheral tumor area, the intermediate vessels increase numerically, up to 54% and the mature ones, up to 18.6%. The peritumoral area was characterized by the absence of immature vessels and the presence of intermediate and mature ones only. It was found the presence of endothelial cell nuclei stained for Ki67 only for immature and intermediate vessels, and never for mature ones. Conclusions: The vascular patterns may contribute to a better stratification of the patient subgroups and antiangiogenic treatment algorithms.

Introduction

Angiogenesis, the process of forming new blood vessels, is evaluated by several methods. Our previous observations indicated the minor involvement of vascular endothelial growth factor (VEGF) in the vessel’s formation associated with infiltrative bladder carcinoma (BC). We found no statistically significant correlation between VEGF expression and vascular microdensity [1]. However, the number of blood vessels, including in the tumor area, was very high compared to normal tissues. Based on these observations, we further investigated the appearance of blood vessels in the tumor and peritumoral areas. The blood vessels in the stroma of malignant tumors are morphologically different from the normal ones, aspect noticed even on the sections stained with conventional morphological methods. The differences in size and trajectory, the thin wall, and the relatively frequent absence of the lumen, without this being an incidence-induced error, attract attention. Nearly two decades ago, Gee et al. (2003) [2] classified tumor-associated blood vessels into immature, intermediate, and mature.

The vessels associated with the mature tumor are structurally similar to the venules, with a wide lumen, endothelial cells with a low proliferation rate, and perivascular cells that may be pericytes or smooth muscle cells. The nuclei of the endothelial cells of the mature vessels are always stained with Hoechst’s reagent. The difference from normal vessels is the messy three-dimensional (3D) architecture and multiple branches. Intermediate vessels have a detectable lumen, but endothelial cells have a higher proliferation rate, which can reach up to 15–20% and do not have perivascular cells, so that in double immunostaining methods express the endothelial marker only. Immature vessels or vascular buds are characteristic of the initial moments of endothelial proliferation. Immature vessels have no detectable lumen, are not perfused, and do not stain with Hoechst’s reagent. The proliferation rate detected with Ki67 is over 50%. The last category, which cannot be included in the classification of tumor vessels, is the leading endothelial cells, which are also known as cells. These cells are observed almost exclusively in the tumor area, among malignant cells, they are fusiform, express endothelial marker and always precede the budding phenomenon observed in immature vessels [3].

Although this classification of malignant tumor-associated blood vessels, already well known and supported by solid arguments, it is rarely discussed in routine pathological diagnosis. Here, we also include the classification systems that have in definition the element V, venula/vein. However, the classification of blood vessel types has been applied in several tumor sites, with promising results. To date, however, there is no publication on infiltrative bladder tumors from this point of view. We investigated this issue in the cases included in the study, hoping to find an answer to the question: which is the optimal target for antiangiogenic therapy, because not all vessels respond to this therapeutic model. Dormant endothelial cells have a long lifespan of hundreds of days. Under these conditions, the identification of divisions in the normal endothelium is exceptional and the nuclear proliferation markers are negative.

In normal tissues, the doubling time of the endothelial cell population is over 1000 days, but it becomes only a few days if the cells are activated. The activation process is signaled by the early expression of endoglin, as shown in several pathological forms of cancer [4, 5, 6]. The first attempts to identify the rate of endothelial cell proliferation in intra-tumoral vessels were not convincing, neither in squamous cell carcinomas (SCCs) of the head and neck nor in malignant melanoma [7].

Aim

The working hypothesis has at least two connotations: one refers to the prognostic value of endothelial proliferation and the other to a potential indicator of the response to antiangiogenic therapy. We consider that these aspects represent a really strong motivation to study the endothelial proliferation in infiltrated bladder tumors, an aspect not reported so far in the literature.

Patients, Materials and Methods

Patients

Fifty-two consecutive cases of T2–T4 infiltrative bladder tumors, aged between 46 and 78 years, were investigated. The main clinical and preclinical parameters were used for diagnostic framing. The used surgical procedure was cystectomy and reconstruction in some cases. The elements of inclusion in the study were represented by the documented existence of the infiltrative tumor and complete data about the extension balance and staging of tumor, node, metastasis (TNM). The samples for the present study were taken only from the radical cystectomy specimen, selecting a representative fragment from the tumor and a fragment from the tumor–bladder wall interface without significant macroscopic changes.

Primary processing

After fixation, samples were dehydrated, clarified, and included in paraffin. The entire procedure was performed in a standardized automatic system with the Thermo Shandon workstation (Thermo Fisher Scientific Inc., Göteborg, Sweden). From each paraffin block were made multiple serial sections with a thickness of 3 μm, adapted to morphological and immunohistochemical (IHC) staining. The morphological evaluation was performed on the sections stained with the Hematoxylin–Eosin (HE) method, by standard technique, with Leica automatic system. The histopathological (HP) type of carcinoma, the degree of differentiation and the level of invasion were established by examination on these sections.

Immunohistochemistry

For the simple and double IHC methods was used Leica Bond-Max autostainer (Leica Biosystems, Newcastle upon Tyne, UK). The unmasking step, with a duration of 20 minutes, was made by using the ER2 solution (distributed by Leica Biosystems, Newcastle Ltd., UK). It was followed by 3% hydrogen peroxide application, five minutes. After this step, in the case of simple immunostainings, the primary antibody – anti-cluster of differentiation (CD)31 [Leica Bond, ready-to-use (RTU), mouse anti-human monoclonal antibody, clone 1A10, 30 minutes incubation time] and Ki67 (Leica Bond, RTU, clone MM1, mouse anti-human monoclonal antibody) were applied. For visualization, we used a peroxidase-based system from the previously mentioned supplier. The incubation time for secondary antibody was eight minutes and for the polymer eight minutes also. As finals steps, 3,3’-Diaminobenzidine (DAB) hydrochloride and Hematoxylin were applied.

In the case of double immunostaining, the following main steps succeeded: peroxidase block (5 minutes), incubation with primary antibody (anti-CD31 for 20 minutes), post primary antibody (eight minutes), polymer (eight minutes), mixed DAB refine (10 minutes), second antibody [anti-smooth muscle actin (SMA) – Leica Bond, RTU, mouse anti-human monoclonal antibody, clone 1A4 and Ki67 – Leica Bond, RTU, mouse anti-human monoclonal antibody, clone MM1 for 20 minutes], components of the alkaline phosphatase-based visualization system: post primary (20 minutes), polymer (20 minutes), red chromogen (10 minutes) and Hematoxylin (five minutes). The stained sections were permanently mounted with Canada balsam conditioner.

Microscopic interpretation and image acquisition

For the evaluation of the stained samples, Zeiss Axiocam 506 (Jena, Germany) and Nikon AY260 microscopes were used. The evaluation of microvascular density (MVD) was performed according to the standard method [8], respectively the choice of three microscopic fields with maximum vascular density, the counting being done at ×200 magnification. The arithmetic mean represented the result; for each slide, the tumor and peritumor area were analyzed. Ki67 was evaluated by the original semi-automatic method [9], on sections with double immunostaining.

Results

Most cases were urothelial carcinomas (n=43), with invasion level T1 in 10 cases, T2 in two cases, and T3 in 31 cases. The degree of differentiation calculated based on the recommendations of the World Health Organization (WHO) Working Group, was predominantly GIII (n=30), characterized by severe cellular anaplasias, areas of variable necrosis, and abundant inflammatory infiltrate. Only two cases had GI and 11 cases, GII. In the conventional form, invasive urothelial carcinoma presented areas, nests, and extensive trabeculae of tumor cells, with or without branches, which dissociate the elements of lamina propria and muscularis propria, depending on the level of invasion. Tumor areas with vascular dilatations, stasis, superficial ulcerations were observed in most cases. Tumor cells showed marked polymorphism, with nuclear atypia and atypical mitosis, the phenomenon of cannibalism being relatively common. As particular HP forms, we mention urothelial carcinoma with clear cells (n=2), characterized by nests and compact areas of tumor cells, apparently well defined, with reduced stroma and tumor cells with clear cytoplasm and severe anaplasia. We observed the sarcomatoid form in two cases, most of the tumor being formed by spindle-shaped cells arranged in bundles that replace all the normal structures of the bladder.

SCC has been identified in four cases. All of these have presented characteristic aspects, respectively the squamous differentiation with many keratosis pearls, the tumor cells being arranged in large, compact, invasive areas. Malignant cells showed severe anaplasia, the cases being classified as GIII (n=3) and GII (n=1). Although with a deep level of invasion in all cases, lympho-vascular invasion was observed by the conventional morphological method in a single specimen.

Adenocarcinoma (ADK) was diagnosed in five cases, all being developed in the bladder dome; proliferation was glandular, similar in some areas even identical to the known aspects of colorectal tumors. These tumors come from the remains of urachus and have associated relatively large areas of the colon-like mucosa (glands with goblet cells and enterocytes). Morphological staining revealed the following distribution of cases (Table 1).

Table 1.

The distribution of cases after the morphological evaluation

Case No.	Histopathological type	T	G	Lympho-vascular invasion
1.	UCC	3A	III	+
2.	SCC	3B	II	-
3.	UCC	3B	III	-
4.	UCC	3B	III	+
5.	UCC	1	II	+
6.	UCC	3A	III	+
7.	UCC	3B	III	-
8.	UCC	1	I	+
9.	UCC	3A	III	-
10.	ADK	3B	III	-
11.	UCC	1	I	-
12.	UCC	1	III	-
13.	UCC	3A	III	+
14.	UCC	3A	III	-
15.	UCC	3B	III	+
16.	UCC	3B	III	-
17.	UCC	2	III	-
18.	UCC	3A	III	+
19.	UCC	1	II	-
20.	UCC	1	II	-
21.	UCC	3B	III	-
22.	UCC	3A	III	+
23.	UCC	3A	II	+
24.	UCC	3A	III	-
25.	UCC	3B	III	+
26.	UCC	3B	III	+
27.	UCC	3B	III	-
28.	UCC	2	III	-
29.	ADK	3B	II	-
30.	ADK	3B	II	-
31.	ADK	3B	II	-
32.	UCC	1	III	-
33.	SCC	3B	III	-
34.	SCC	3B	III	-
35.	UCC	1	II	-
36.	UCC	1	II	-
37.	UCC	3B	II	+
38.	UCC	3B	II	-
39.	UCC	3B	II	+
40.	UCC	3A	III	-
41.	UCC	3B	III	+
42.	UCC	3B	III	+
43.	UCC	3B	III	-
44.	UCC	3B	III	+
45.	UCC	3B	III	+
46.	UCC	3B	III	-
47.	ADK	2	II	-
48.	SCC	3B	III	+
49.	UCC	3A	III	-
50.	UCC	3A	II	+
51.	UCC	3B	III	+
52.	UCC	1/Ta	II	-

ADK: Adenocarcinoma of the urinary bladder; G: Grade; SCC: Squamous cell carcinoma of the urinary bladder; T: Tumor; UCC: Urothelial cell carcinoma

Vessel types associated with infiltrative bladder tumor were evaluated by the double IHC staining method, based on a cytoplasmic endothelial marker (CD31, final reaction product visualized in brown) and a specific marker for contractile filaments found in perivascular cells (smooth muscle actin, red reaction final product). For immature and intermediate vessels, only the endothelial marker was evident (Figure 1A, 1B).

Figure 1.

(a) Immature vessels with lumen, but without perivascular cells; (b) Immature and intermediate vessels; (c) Mature vessels showing expression for endothelial and perivascular markers (endothelial cells stained in brown, smooth muscle cells stained in red); (d) Mature vessel with dormant endothelium, Ki67-positive cells in the perivascular cell only; (e) Reduced endothelial cell proliferation, Ki67-positive endothelial cell; (f) Rapid endothelial cell proliferation; (g) Mature vessels with dormant endothelium; (h) Immature vessels with moderate proliferative endothelium; (i) Immature vessels with Ki67 signals present at nuclear level; most of the tumor cells are positive for Ki67. CD31/SMA double immunostaining: (a) ×400; (b and c) ×100. Ki67 immunostaining: (d–f) ×400. CD31 (red)/Ki67 (brown) immunostaining: (g and h) ×200; (i) ×400. CD31: Cluster of differentiation 31; SMA: Smooth muscle actin

In all the infiltrative bladder tumors studied, we identified all three types of vessels associated with the tumor, as described above. Although the density of microvessels was similar as a result with the MVD values obtained on the CD31 simple immunostaining. On CD31/SMA double immunostaining, we noticed that immature vessels are most numerous in the tumor area in the central area, where they represent 62.25% of all positive structures for the endothelial marker. In this area, intermediate vessels represent 35.1%, and mature vessels only 6.65%. The prognostic value of MVD increases significantly if this method of quantification is applied, which excludes virtually all mature vessels.

The ratio between the types of vessels changes in the peripheral tumor area, where the intermediate vessels increase numerically, up to 54% and the mature ones, up to 18.6%. In the peritumoral area, we practically did not observe immature vessels, but only intermediate and mature ones (Figure 1C).

CD31/SMA double immunoreaction analysis revealed a significant correlation between the number of immature, intermediate and mature vessels from the intratumoral, peritumoral areas and G (p=0.0001) of all evaluated cases. No significant correlation was found between G and lympho-vascular invasion in all evaluated cases (p=0.5954). A significant correlation was noticed between the number of immature, intermediate and mature vessels and lympho-vascular invasion in the intratumoral area (p=0.0070; p=0.0077; p=0.0338) and peritumoral area (p=0.0001; p=0.0078; p=0.0419). A significant correlation was found between the number of immature, intermediate and mature vessels from the intratumoral and peritumoral area for all evaluated cases (p=0.0001).

In the cases of urothelial carcinoma, a significant correlation was noticed between immature, intermediate and mature vessels from the tumor’s inner area, peritumoral zone, and grade score (p=0.0001). Representatives’ correlations were found between the intratumoral immature, intermediate and mature vessels, and lympho-vascular invasion (p=0.0344; p=0.0330; p=0.0343) and between the number of peritumoral intermediate and mature vessels and lympho-vascular invasion (p=0.0323; p=0.0389).

SCC type was characterized by a significant correlation between the number of intratumoral and peritumoral immature, intermediate and mature vessels (p=0.0001; p=0.0197; p=0.0001). Significant correlations were noticed between the number of peritumoral immature vessels and G and lympho-vascular invasion also (p=0.0001).

Significant correlations identified in the ADK type were between the number of intratumoral immature, intermediate and mature vessels and their counterparts in the peritumoral areas (p=0.0001); between G and lympho-vascular invasion (p=0.001); between the number of intratumoral mature vessels and G (p=0.0068); between the number of peritumoral immature and mature vessels and G (p=0.0001 and p=0.0053).

The interrelations between the type of vessels and the HP type in the intratumoral and peritumoral areas are summarized in Figure 2.

Figure 2.

The interrelation between MVD, type of vessels and histopathological type in the intratumoral and peritumoral areas. ADK: Adenocarcinoma of the urinary bladder; Immature itu: MVD of immature vessels in the intratumoral area; Immature peri: MVD of immature-type vessels in the peritumoral area; Intermediate itu: MVD of intermediate-type vessels in the intratumoral area; Intermediate peri: MVD of intermediate-type vessels in the peritumoral area; Mature itu: MVD of mature-type vessels in the intratumoral area; Mature peri: MVD of mature-type vessels in peritumoral area; MVD: Microvascular density; SCC: Squamous cell carcinoma of the urinary bladder; UCC: Urothelial cell carcinoma

Staining of sections with Ki67 reveals several aspects in the blood vessel wall. In mature vessels with dormant endothelium, no brown-stained nuclear signals are observed in the endothelium, but only in the perivascular layer, as in normal conditions (Figure 1D). Endothelial proliferation was of two major types: low and intense, respectively. At slow or low-density proliferation, only 1–2 endothelial cells showed a positive reaction at the nuclear level (Figure 1E). In the latter case, most endothelial cells were stained for Ki67 (Figure 1F).

Quantifying endothelial proliferation on the double immunoreaction for CD31 and Ki67 is a real challenge for the examiner, because in infiltrative bladder tumors a large number of malignant cells are also positive at the nuclear level. Therefore, we considered only the cells that express Ki67 at the nuclear level and co-express in red CD31. A particular aspect that we report was the absence of the Ki67-positive endothelial cell nuclei in the mature vessels (Figure 1G) and the presence of endothelial cell nuclei stained for Ki67 in the immature and intermediate vessels (Figure 1H, 1I).

Discussions

MVD is the most commonly used parameter for characterizing angiogenesis. The approach to angiogenesis only by this measurement is largely false, because the only information it brings is the number of blood vessels at a time in a particular area of the organ or lesion. The results are often controversial in terms of clinical relevance and prognosis. Activation of endothelial cells is an important moment for the development of vessels starting from previous ones. Currently, seems that the most useful marker in this regard is endoglin (CD105), which can be detected at the protein level. Endoglin is positive in activated endothelial cells, the newly formed vessels having large perivascular spaces. Endoglin is a surface homodimeric component of type I transforming growth factor-beta (TGF-β). CD105 expression signals activation and usually precedes proliferation. For the reasons stated above, we consider that other methods are needed to indicate the angiogenic potential on the one hand and the response to therapy on the other. Microscopic observations on malignant tumors have highlighted the different structure of the vessels in the tumor area compared to normal tissues, already many years ago. Forty years ago, nine categories of blood-type vessels identified in the tumor were proposed. Due to the remarkable anticipation of the results obtained with methods introduced much later, we present below this first attempt to define the vessels in the tumor area in terms of their morphological characteristics.

Warren classification of tumor vessels (1979) [10] contains in the first category arteries and smallest arteries – arterioles as main types; the second category and third category included capillary types of vessels and capillary buds; fourth category: sinusoids; fifth category consists of channels not covered by endothelium; sixth category represented by large capillary type of vessels; seventh category: fenestrated capillary type of vessels; eight category included veins and their smallest version – venules; ninth category was represented by special vascular dispositive such as arteriovenous type of anastomoses.

There are a number of major differences between normal and intratumoral blood vessels. Intratumoral vessels often have an incompletely mature wall and, as a result, the architecture is abnormal. The collapse of the inner tumor zone vessels often occurs. The main consequences are hypoxia and necrosis. Thus, the development of tumor cells can no longer be supported. The anti-CD105 antibody is useful for differentiating between normal vessels and those formed by tumor angiogenesis, because the last one is CD105 positive.

Gee et al. (2003) demonstrated that tumor vessels can be segregated into three categories, based on size, perfusion, endothelial cell proliferation and the presence of pericytes: unperfused buds of (immature), intermediate and mature endothelial cells [2]. Immature vessels consisting of buds of endothelial cells that form from the wall of functional vessels, are not perfused, and have a high proliferation rate. The buds may be isolated endothelial cells, with no visible lumen, positive for CD31 and von Willebrand factor (vWF). These cells are negative for the S100 protein and represent between 18% and 25% of the total intratumoral vascular structures in large tumors and predominate numerically in small tumors. Isolated endothelial cells are not infused and do not stain with Hoechst 33342 reagent, which marks the nuclei of endothelial cells exposed to blood flow. The proliferation rate demonstrated with proliferating cell nuclear antigen (PCNA) is between 39% and 69%. In our study, we noticed the main three types of vessels in the tumor bladder: immature, intermediate and mature types.

The intermediate vessels are small, infused, which do not have pericytes or smooth muscle cells in the constitution of the wall. They are infused vessels, in which over 90% of endothelial cells are stained with Hoechst 33342 reagent. The incidence of this type of intratumoral vessel varies between 33% and 43%. Mature vessels are large, have a wide lumen and thin wall, endothelial cells are dormant, have minimal proliferative potential, contain pericytes in the wall and rarely form endothelial buds. The proliferation rate is low, the index for PCNA being between 12% and 18%. These vessels branch repeatedly and give rise to the intermediate ones, without pericytes.

Tumor vessels are characterized by marked heterogeneity and surprisingly, their nature is quite little known in terms of human malignancies. The rate of endothelial cell proliferation differs from organ to organ and does not always follow the rules set out above. Thus, the overall proliferation rate of endothelial cells in human tumors is 0.15% in prostate cancer, 2.2–2.7% in breast cancer, and 9.9% in colorectal cancer. These data were confirmed on six different types of human tumors, in which similar values were reported, the reactions for Ki67 and PCNA being at the limit of detection in endothelial cells in normal tissues. These data correspond to the average of all three types of vessels in the Gee classification and do not necessarily correlate with MVD [11, 12]. In fact, in renal, colic, mammary, pulmonary carcinomas and glioblastomas, MVD has lower values in the tumor area than in the corresponding normal tissue [13]. Under these conditions, the endothelial cell proliferation rate can be considered the most realistic parameter for quantifying angiogenesis. The three types of intratumoral vessels can be highlighted by performing double immunostaining for CD31 (or CD34) and smooth muscle actin, the latter being positive for both smooth muscle cells in the vascular wall and pericytes. Most vessels in growing tumors are immature and intermediate. Anti-angiogenic therapy drastically reduces the number of these vessels, while mature ones do not change significantly. The degree of maturation of neoformation vessels correlates with the response to angiogenesis inhibitor therapy. This is demonstrated by renal cell carcinomas, which, although they express VEGF in over 70% of cases, do not respond significantly to anti-vascular therapy, as most vessels are mature. The classification of intratumoral vessels in these categories indicates that the therapeutic target is represented by immature and intermediate vessels, a demonstrated and experimental aspect in animals with tumors treated with interleukin-12 (IL-12) [14].

Some data demonstrated the interrelation between VEGF-A-mediated delivery from vWF of endothelial cells and BC progression. Therapeutic protocols that stopped the previously mentioned mechanism may decrease the tumor-associated hypercoagulation and metastatic tendency. The evaluation of vessels number in healthy tissue, in low- and high-grade tumors showed a high number of activated blood vessels in the peritumoral area than in healthy tissue, and in high-grade compared to low-grade tumors [15]. In our study, proliferative endothelial cells were noticed in the immature and intermediate vessels, mainly in the central part of tumor area.

The classic antiangiogenic drugs, such as Bevacizumab, Sunitinib, Sorafenib, Aflibercept, Pazopanib and Vandetanib, but also novel antiangiogenic therapy (Ramucirumab, Icrucumab, Regorafenib, Nintedanib) have been tested in clinical trials for BC [16]. No compound has yet proven significant single-agent efficacy. This may be the reason why no antiangiogenic therapy was included in the BC therapeutic protocols until this moment. It was noticed that subcohorts had favorable results compared to extensive clinical trials. The observation of our study may be useful for the better classification of the patients on one hand and for proper understanding of treatment algorithms used in combination therapy on the other hand.

Conclusions

In this paper, we demonstrated the existence of immature and intermediate vessels in infiltrative bladder tumors. Immature vessels predominate intratumorally and intermediate ones intratumorally peripherally and peritumorally. Only these two types of vessels can be a viable therapeutic target for anti-vascular medication. MVD calculated only for immature and intermediate vessels has a high prognostic value. The endothelial cell proliferation rate accurately reflects the activation and progression of tumor angiogenesis and may be a major candidate for evaluating the efficacy of anti-vascular and antiangiogenic therapy.

Conflict of interests

The authors declare that they have no conflict of interests.