Intratumoral Recombinant Human Interleukin-12 Administration in Head and Neck Squamous Cell Carcinoma Patients Modifies Locoregional Lymph Node Architecture and Induces Natural Killer Cell Infiltration in the Primary Tumor

Head and neck squamous cell carcinoma (HNSCC) provides an excellent model to study the effects of locoregional immunotherapy with cytokines. HNSCC is characterized by the occurrence of regional lymph node metastases. Therefore, resection of the primary tumor, along with a regional lymph node dissection, is part of the standard treatment in patients with limited disease. HNSCC is an immunogenic tumor, as shown by a variable amount of infiltrating lymphocytes and other immune cells. In principle, these infiltrating leukocytes are susceptible to activation by immunomodulatory cytokines (1). However, several types of immune dysfunction, starting at the site of the tumor and eventually generalized, have been reported (2, 3). Locoregional immunotherapy given preoperatively allows for study of the effects of the treatment on the primary tumor, the lymph nodes, and peripheral blood. Tumors in the oral cavity or oropharynx carcinoma permit local injection of cytokines.

Interleukin-12 (IL-12) is a heterodimeric cytokine that consists of two disulfide-linked subunits (i.e., IL-12p40 and IL-12p35; ref. 4). IL-12 has a wide range of biological activities (4, 5) and has effect on both the innate and adaptive immune system. IL-12 stimulates the proliferation and activation of CTLs and natural killer (NK) cells and induces the production of cytokines, especially IFN-γ (6–8). IL-12 is the key cytokine in the induction of T helper 1 (Th1) responses and thereby of cellular immunity (9). In addition to a crucial role in initiating cell-mediated immune responses, IL-12 also has an effect on humoral immunity. IL-12 stimulates B-cell growth by inducing IFN-γ production in B cells in vitro (10), causes immunoglobulin isotype selection (11), and triggers a cascade of molecular events in human B cells similar to Th1 commitment (12). Furthermore, IL-12 inhibits angiogenesis (13, 14). IL-12 has antineoplastic activity in experimental murine tumor models (15, 16). Several phase I (17–22) and phase II studies in various cancer types (23–26) have been done with either i.v., s.c., or i.t. administration of recombinant human IL-12 (rhIL-12). However, only a limited number of clinical responses were observed, probably because of opposing immune mechanisms.

We have done a phase II study of the administration of rhIL-12 to HNSCC patients before surgery to evaluate the immunologic effects on the primary tumor and regional lymph nodes (26). Toxicity, existing of increase of liver transaminases, fatigue, and metabolic acidosis, was higher than expected at the dose levels used. Dose-dependent plasma IFN-γ and IL-10 increments were detected. In peripheral blood, a rapid, transient reduction in lymphocytes, especially NK cells, was observed. The redistribution of lymphocytes to lymphoid organs led to enlarged lymph nodes. Real-time quantitative PCR analysis of blood mononuclear cells in the lymph nodes showed a 128-fold increase of IFN-γ mRNA. A switch from the Th2 to a Th1 profile in the lymph nodes related to the peripheral blood occurred in the IL-12-treated patients.

The objective of this paper was to unravel the cascade of effects of IL-12 on the cells of the immune system in the primary tumor and the locoregional lymph nodes. The aim was to determine which cells seem to be most important in mediating the antitumor effects of IL-12 in humans. Beneath this, effects on other cells of the innate or adaptive immune system (e.g., the dendritic cells, DC), B cells, eosinophils, macrophages, and neutrophils, were studied. We report on the histologic and immunohistochemical findings in the primary tumor and lymph nodes of the IL-12-treated patients compared with control patients.

Patient Selection. All patients had histologic proof of HNSCC, with the primary tumor in the oral cavity or oropharynx, staged as T1-4, N0-2, and M0, for which surgical resection including a supraomohyoid or radical neck lymph node dissection was planned. Patients were previously untreated (i.e., they did not have prior surgery, radiotherapy, or any systemic therapy). The tumor, with a diameter not exceeding 5 cm, had to be accessible for local injection. Further eligibility criteria included age between 18 and 75 years, WHO performance score of 0 to 2, life expectancy of >3 months, adequate renal function (serum creatinine ≤1.5 times normal), adequate hepatic function (serum bilirubin ≤1.5 times normal; alanine aminotransferase and aspartate aminotransferase ≤2 times normal), normal serum calcium (≤11 mg/dl), serum hemoglobin ≥9 g/dl, granulocytes ≥1,500/μL, and platelets ≥100,000/μL. Systemic corticosteroids were not allowed. Patients with major concurrent disease were excluded, as were patients known to be positive for HIV or hepatitis B surface antigen.

To allow comparison of the histologic and immunohistochemical variables of the 10 rhIL-12-treated HNSCC patients, we collected also primary tumor resection material and lymph nodes from 20 control patients; these patients were eligible for the study but preferred not to receive the rhIL-12 injections.

The local regulatory committee approved the study. All patients gave written informed consent.

Study Design and Treatment Schedule. In this single-center, open-label, nonrandomized, phase II study, rhIL-12 was supplied by Wyeth (Cambridge, MA) and was given at two dose levels of 100 and 300 ng/kg, by single or multiple i.t. injections in the primary tumor only. Five patients per dose level were planned. Patients were treated once weekly in the normal waiting period before surgery, with a minimum of two and a maximum of four doses. Surgery was never postponed because of participation in this study. The last planned injection was given 24 hours before surgery. This was based on the results of the prior phase Ib study (24), in which the most pronounced immunologic effects in peripheral blood were seen after 12 to 24 hours. The first injection was given as inpatient treatment; others were on an outpatient basis, with an observation period of 1 hour after each injection. The mean volume of the injected rhIL-12 was 0.51 mL (range, 0.37-1.1 mL), related to the dose level and the weight of the patient.

Instead of five planned patients per dose level, six patients received 100 ng/kg and four patients 300 ng/kg, because of toxicity. Two patients at the 300 ng/kg level received only one injection, 8 and 15 days before surgery, respectively. One patient received 300 ng/kg as first administration and 100 ng/kg as second administration (26). Thus, only one patient received the planned dosages of 300 ng/kg. Therefore, it was not possible to separate the 100 and 300 ng/kg dose level in this study.

Handling of the Resected Material and Tissue Preparation. Immediately after resection of the primary tumor and the lymph nodes, the material was put on ice. The lymph node specimen was cut out freshly by the pathologist (PdW). The neck was divided into six lymph node regions (I-VI), from which all lymph nodes were collected. Dependent on the size of the lymph node, part of the lymph node was directly snap frozen. If the primary tumor was sufficiently large, a sample was taken for direct snap freezing. Thereafter, the primary tumor was fixated in 4% (v/v) phosphate buffered formalin and cut out on the following day, or after decalcification. The lymph nodes were fixated in unifix, dehydrated, and embedded in paraffin. Tissue sections of 4 μm were cut, mounted onto glass slides pretreated with 2% 3-aminopropyltriethoxysilane (Sigma, St. Louis, MO), and dried overnight. Serial sections were stained with H&E or processed for immunohistochemistry.

Scoring of H&E Sections and Categories. Routinely H&E-stained histopathologic sections were used to determine the differentiation (WHO and Broders), growth pattern, and inflammatory reaction of the primary tumor.

Routinely, H&E-stained histopathologic sections of the lymph nodes were scored as follows: (i) primary follicles and (ii) secondary follicles: 0, no follicles; 1, very few follicles; 2, few follicles; 3, many follicles; 4, filled with follicles; (iii) paracortical hyperplasia: 0, absent; 1, small area; 2, large areas; (iv) sinus histiocytosis: 0, absent; 1, very small and few areas; 2, small and multiple areas; 3, “bands” that occupy <40% of the lymph node; 4, bands that occupy ≥40% of the lymph node.

Immunohistochemical Staining. All primary tumors were immunohistochemically stained with the antibodies listed below. One lymph node was stained from every IL-12-treated patient and from 10 control patients. The 10 control patients were selected based on patient and tumor characteristics and were comparable with the 10 IL-12-treated patients. The lymph nodes of both groups were selected based on histologic features and size. The average size, number of primary and secondary follicles, and degree of paracortical hyperplasia and sinushistiocytosis were identical in H&E-stained specimens and immunohistochemical-stained specimens in each group. All chosen lymph nodes were without metastasis.

Tissue sections were incubated by the following primary antibodies: anti-CD3 (clone SP7, 1:100, Lab Vision-Neomarkers, Immunologic, Duiven, the Netherlands), anti-CD4 (clone 1F6, 1:80, Lab Vision-Neomarkers, Immunologic), anti-CD8 (clone C8/144B, 1:80, DakoCytomation, Heverlee, Belgium), anti-CD20 (clone L26, 1:500, DakoCytomation), anti-CD21 (clone 1F8, 1:200, DakoCytomation), anti-CD56 (clone 123C3, 1:10, Monosan/Sanbio, Uden, the Netherlands), anti-CD57 (clone NC1, 1:5, Immunotech/Coulter, Mijdrecht, the Netherlands), anti-CD68 (clone Kp1, 1:2,000, DakoCytomation), anti-CD79a (clone JCB117, 1:500, DakoCytomation), BerMacDRC (clone BerMac, 1:20, DakoCytomation), DC-SIGN/anti-CD209 (20 μg/mL, AZN-D1; ref. 27), dendritic cell lysosyme-associated membrane glycoprotein (DC-LAMP)/anti-CD208 (clone 104.G4, 1:10, Immunotech/Coulter), Langerin/anti-CD207 (clone DCGM4, 1:40, Immunotech, Marseille, France), anti-CD11c (clone SHCL-3, 1:5, BD PharMingen, San Diego, CA), intercellular adhesion molecule-2 (ICAM-2)/anti-CD102 (CBR-IC2/2; ref. 28), neutrophil elastase (1:100, DakoCytomation), AA-1 (clone AA1, 1:400, DakoCytomation), EG-2 (1:20, Monosan/Sanbio), and MIB (Clone Mib-1, 1:200, DakoCytomation).

In brief, sections were deparaffinized in xylene and rehydrated. Endogenous peroxidase was blocked with 0.3% hydrogen peroxide in methanol for 20 minutes. Antigen retrieval was done using 0.1 mol/L citrate buffer pH 6.0 (anti-CD3, anti-CD8, anti-CD20, anti-CD56, anti-CD57, anti-CD79a, anti-CD207, and MIB), microwave cooking in 0.1 m EDTA pH 8.0 (anti-CD4, anti-CD209, anti-CD11c, and anti-CD102), and fresh Pronase 0.1%/PBS (anti-CD21, BerMacDRC, anti-CD208, neutrophil elastase, AA-1, and EG-2), respectively. Subsequently, sections were preincubated with 1% bovine serum albumin. After overnight incubation with the primary antibody, the secondary biotin-conjugated antibody, and a tertiary complex of streptavidin-avidin-biotin conjugated to 3-amino-9-ethyl-carbazole were applied. Finally, the sections were counterstained with Mayer's hematoxylin. Incubation with PBS instead of the primary antibody served as a negative control.

Semiquantitative Scoring of Immunohistochemical Staining of the Primary Tumors. The scoring of the markers was done within the tumor (intratumoral infiltrating cells, present in the stroma of the tumor), as well as separately along the invasive border of the tumor in the surrounding tissue (peritumoral infiltrating cells). If equal expression was observed both intratumorally and peritumorally, or if cryosections were scored, only “overall” was scored. After analysis of the distribution of the numbers of positive cells, the infiltrate was scored semiquantitatively, using three categories: 0, none/few; 1, moderate; 2, many/extensive or five categories: 0, none; 1, few; 2, moderate; 3, many; 4, extensive.

Quantitative Analysis of Immunohistochemical Staining of the Lymph Nodes. Quantification of the immunohistochemical staining product was done in stored images of completely scanned tissue sections. Images were acquired with an AxioCam MRc (Carl Zeiss, Oberkochen, Germany) connected to an AxioPlan 2 Imaging microscope (Carl Zeiss). The microscope was equipped with a computer-controlled scanning stage (eight specimen stage, Märzhäuser GmbH, Wetzlar, Germany, controlled by a Ludl MAC5000 controller, Ludl Electronic Products Ltd., Hawthorne, NY). Images were acquired using a 10× objective (Plan Apochromat, NA = 0.32; specimen level pixel size 1.06 × 1.06 μm²). Images were corrected for unequal illumination using a stored shading reference image. Each microscopic field was individually autofocused before acquisition. All image acquisition and processing was done using custom written macros in KS400 image analysis software (version 3.0, Carl Zeiss).

For each specimen, the region of interest was interactively defined using a live image display. Images of consecutive individual microscopic fields were acquired and stitched together into large 24-bit RGB TIFF images. Automatic detection of the region occupied by tissue was done in each stored composite image, based on automatic detection of the background peak in the histogram of the red camera channel. In a few cases with artifacts in the background, the automatically detected threshold was not adequate and therefore was interactively determined. Also, in a number of cases parts of the background were interactively labeled to be excluded from analysis, to remove small artifacts that were incorrectly thresholded as tissue.

To recognize positively stained regions, the ratio between the red and green pixel intensity was thresholded. The fast red staining product absorbs light in the green part of the spectrum whereas mostly transmitting red light. All other components present in the specimens either absorb red light or transmit approximately equal amounts of light in the red and green parts of the spectrum. Therefore, the ratio between the red and green pixel intensity is maximum for the red dye, with low values for all other tissue components. For each specimen, the threshold was interactively set. Finally, the area of the tissue and the area occupied by positive immunostaining were calculated. The system is calibrated so that all variables are expressed in micrometers. The automated procedure was used to quantify tissue sections with monoclonal antibody anti-CD3, anti-CD4, anti-CD20, anti-CD79a, anti-CD68, anti-DC-SIGN, anti-CD11c, and anti-ICAM-2. For those cases, the threshold for recognition of positive staining was manually set by two independent observers individually (CvH/JvdL), to study the observer reproducibility. For the other markers, the positive area was not considered appropriate scoring. In these cases, the number of positive cells was counted interactively using the ks400 image analysis system described above. Positive cells were manually marked on the image screen by mouse clicks. The computer gave the number of positive cells. Also, the area of the region of interest of the lymph node was determined by interactive drawing on the computer screen.

Sorting of Lymph Node Suspensions. Lymph node suspensions were obtained as described previously (1) and stored in liquid nitrogen until further analysis. Separate populations of CD4-, CD8-, and CD56-positve cells were obtained via fluorescence-activated flow cytometry. To this extend, lymph node suspensions were thawed and washed with DMEM containing 10% FCS and PBS containing 0.5% (w/v) bovine serum albumin, respectively. Suspensions were divided in two portions and either incubated with FITC-conjugated anti-CD4 (clone 13B8.2, Immunotech) and phycoerythrin-conjugated anti-CD8-FITC (clone B9.11 Immunotech) and anti-CD56-phycoerythrin (clone NCAM16.2, BD Biosciences, San José, CA) for 10 minutes at room temperature. After incubation, cells were washed with PBS/0.5% bovine serum albumin and sorted on an Epics Ultra HyperSorter flow cytometer (Beckman Coulter, Miami, FL). Cells (3,000-100,000) were captured in cold PBS/0.5% bovine serum albumin. After sorting, cells were pelleted (300 G, 5 minutes, 4°C) and stored at −80°C until further analysis.

Analysis of IFNγ Expression. Analysis of IFN-γ expression was done in five IL-12-treated and five control patients. These patients had comparable patient characteristics. RNA from the sorted cell populations was isolated using the RNeasy Kit (QIAGEN Sciences, Germantown, MD) according to manufacturer's instructions and residual DNA was digested. After digestion, RNA cleanup was done using the RNeasy kit according to the manufacturer's instructions. RNA isolated from 3,000 cells was reversed transcribed with SuperScript-II (Invitrogen, San Diego, CA) according to the manufacturer's instructions in a total volume of 20 μL. PCR was done on 1 μL cDNA with ThermoPerfect (Integro, Denver, CA) according to standard procedures using a primer pair for IFN-γ (5′-TGACCAGAGCCAAAAGA and 5′-CTCTTCGACTCTGAAACAGC; 1.25 mmol/L MgCl₂; T_m, 60°C) or for the low copy peptidylpropyl isomerase, a housekeeping gene product (PPIA, 5′-AGGTCCCAAAGACAGCAGAA and 5′-GTCTTGGCAGTGCAGATGAA; 2.5 mmol/L MgCl₂; T_m, 63°C) that served as control for equal input.

Statistics. The clinicopathologic characteristics of the IL-12-treated and control patients were analyzed by the use of a contingency table; statistical significance was evaluated using the Fisher exact test. Nonparametric tests according to the Mann-Whitney U test and Wilcoxon signed-rank test were done to compare two independent and two dependent samples, respectively. Differences in proportions were evaluated by means of the χ² test or Fisher exact test. For correlation analysis between variables, Spearman's correlation coefficient was calculated. For all tests, P < 0.05 was considered statistically significant. All statistical analysis was done with two-tailed tests using SPSS 11.0 for Windows.

Patient Characteristics and Histologic Features of the Primary Tumors. Ten patients (4 men and 6 women) with a median age of 55 years (range, 46-69) were enrolled in the study (see Table 1). The median WHO performance score was 1 (range, 0-1). Tumor sites were the oral cavity (n = 8) and oropharynx (n = 2). The T and N tumor stages are summarized in Table 1. Radical modified neck dissections and supraomohyoidal neck dissections were done in eight and two patients, respectively. The characteristics of the IL-12-treated patients were not significant different from the 20 control patients (Table 1).

In one IL-12-treated patient, the primary tumor could not be examined because no tumor was present in the resected area. The biopsy taken to prove the diagnosis showed squamous cell carcinoma; most likely, this was an excision biopsy. The lymph nodes of this patient were scored.

No obvious differences were seen in the differentiation and growth pattern of the primary tumors, or infiltration of cells of the immune system in the primary tumors of the IL-12-treated (n = 9) versus the control patients (n = 20) detected on H&E-stained tissue sections (Table 2).

Immunohistochemical Differences in the Tumor-Infiltrating Immune Cells. An increase in the CD56⁺ NK cells was seen in the tumor in the IL-12-treated patients (P < 0.01; Table 3; Fig. 1). In one IL-12-treated and one control patient, staining of the primary tumor was false negative due to decalcification of the resected material. Six of the eight IL-12-treated patients and three of the 19 control patients had a high expression of CD56 (scoring of 3 or 4). The patients, irrespectively if they were IL-12-treated or not, with a high expression of CD56⁺ cells had a better prognosis with a longer overall survival than the patients with a low expression of CD56⁺ cells (scoring of 0, 1, or 2; P < 0.01; Fig. 2). On the other hand, no difference was seen in the CD57⁺ (NK and subset of T cells) cells. Peritumoral CD20⁺ B cell infiltration was very extensive in some IL-12-treated patients. However, the increase in peritumoral B cells was not significantly different between the IL-12-treated and control patients (P = 0.06; Table 3; Fig. 1). The same trend was seen for the peritumoral CD79a⁺ cells (B cells and plasma cells). Intratumorally no differences were seen in CD20⁺ and CD79a⁺ cells. The infiltration of CD3⁺ (T cells), CD4⁺ (T helper cells), and CD8⁺ (T cytotoxic cells) was similar in both groups.

A remarkable finding was the altered distribution of the CD68⁺ (macrophages) cells in IL-12-treated patients. In all IL-12-treated patients the majority of the CD68⁺ cells were located peritumorally (P < 0.05; Table 3), whereas in controls the CD68⁺ cells were more equally spread. No differences between the total numbers of intratumoral and peritumoral CD68⁺ cells were seen.

Next we analyzed the presence of infiltrating dendritic cells (DCs), using novel DC markers. In both groups the DC-SIGN⁺ cells were located in the stroma very near to the tumor nests, and the DC-LAMP⁺ cells were located deeper in the stroma, with more distance to the tumor cell nests (data not shown). No differences were detected in the numbers of DRC⁺ (follicular dendritic cells), DC-SIGN⁺, DC-LAMP⁺, Langerin⁺, and CD11c⁺ cells. Also, ICAM-2, one of the ligands of DC-SIGN, was not different between the groups.

A somewhat lower number of eosinophils was observed in the IL-12-treated patients, but this difference was not statistically significant in the entire group of patients (P = 0.06; Table 3). No differences were detected in the number of mast cells. Whereas most intratumoral immune cells were located in the stroma between the tumor nests, neutrophils were located in the stroma and also within the tumor cell nests. Although the numbers of peritumoral and intratumoral (in the stroma) neutrophils were not different between the two groups, a trend to a lower number of neutrophils in the tumor nests was observed in the IL-12-treated group (Fig. 1).

Histologic Changes in the Lymph Nodes. In the lymph node dissection specimen, the mean diameters of the resected lymph nodes were 7.9 mm (range, 7-8.7 mm) in the IL-12-treated group and 6.1 mm (range, 4.6-8.5 mm) in the control group (P < 0.001). In patients who underwent a radical lymph node neck dissection, the mean number of the lymph nodes was 42 (range, 31-50) in the IL-12 group (n = 7) versus 28 (range, 15-42) in the control group (n = 14; P < 0.005; Fig. 3).

In total, 336 lymph nodes in 10 IL-12-treated and 469 lymph nodes in 20 control patients were collected and scored per patient for the presence of primary and secondary follicles, paracortical hyperplasia, and sinus histiocytosis. The number of secondary follicles, primary follicles, and the presence of paracortical hyperplasia were strongly decreased in the IL-12-treated patients compared with the control patients (P < 0.005, P < 0.005, and P < 0.05, respectively; Fig. 4). Consequently, the lymph nodes had a much more monotonous appearance in IL-12-treated patients (Fig. 5). In the IL-12-treated patients, the secondary follicles contained smaller germinal centers. Furthermore, the outline of the primary and secondary follicles within the lymph node was less sharp and seemed less separated from the paracortex. The degree of sinus histiocytosis was similar in both groups.

Immunohistochemical Differences and mRNA IFN-γ Expression in the Lymph Nodes. To investigate the presence and distribution of leucocytes within the lymph nodes, quantitative analysis after their immunohistochemical detection was done (Table 4; Fig. 6).

The CD20⁺ B cells showed a different distribution in IL-12-treated patients. The B cells were not only located in the follicles but were also situated outside the follicles (Fig. 6). However, the percentage area of CD20⁺ cells was not different between the lymph nodes of the IL-12-treated and control patients. The immunohistochemical analysis did not reveal a major difference in the number of CD56⁺ NK cells (106/mm² versus 78/mm²). The number and distribution of CD3⁺, CD4⁺, CD8⁺, and CD57⁺ (outside the germinal centers) cells were not different between the IL-12-treated and control patients. Reverse transcription-PCR of isolated cell suspensions of the lymph nodes showed that the CD56⁺ cells were the key IFN-γ producing cells in the IL-12-treated patients. Also, the IFN-γ mRNA expression in CD8⁺ and CD4⁺ cells increased compared with control patients (Fig. 7). Thus, although the number of CD56⁺, CD8⁺, and CD4⁺ cells did not increase, IL-12 treatment resulted in their activation.

Because dendritic cells and macrophages are the main producers of IL-12, we also investigated the effects of the given IL-12 on these cell populations. IL-12 induced an impressive decrease in the number of DC-LAMP⁺ cells (mature DCs) in the paracortex (P < 0.005; Table 4; Fig. 6). The localization of DC-LAMP⁺ cells corresponded with the paracortical hyperplasia seen with H&E staining. Thus, the decrease of DC-LAMP⁺ cells correlated with the reduction of paracortical hyperplasia in IL-12-treated patients. Although the number of DC-LAMP⁺ cells was diminished, no differences were seen in the other DC or macrophage markers [DC-SIGN (immature DCs), CD11c (DCs), Langerin (Langerhans' cells), CD21 (follicular DCs), and CD68 (macrophages)]. In both IL-12-treated and control patients, DC-SIGN was expressedmainly in sinus histiocytosis and to a lesser degree in the paracortex. The stromal expression of ICAM-2, especially in sinus histiocytosis, was lower in the IL-12-treated patients (P < 0.001; Table 4; Fig. 6), but the expression on the endothelium lining the vessel walls was similar.

More neutrophils, especially around regions of sinus histiocytosis, and fewer mast cells were present in the IL-12-treated patients (P < 0.01 and <0.005, respectively; Table 4; Fig. 6). The number of eosinophils was not different between the groups. However, in the IL-12-treated patients a negative correlation between the number of neutrophils and eosinophils was noticed (r = −0.82; P < 0.005).

The main purpose of this histopathologic and immunohistologic study was to evaluate the effects of IL-12 treatment on the immune cells in the primary tumor and lymph nodes in patients with HNSCC. In the primary tumor, the number of CD56⁺ NK cells was increased in IL-12-treated patients compared with control patients. Partly in contrast with the previously described flow cytometry data (26), which showed a significant increase of CD3⁻CD16⁺CD56⁺ NK cells (1.6% versus 0.7%; P < 0.001) after IL-12 treatment, with immunohistochemical analysis we did detect the same trend in the increased number of CD56⁺ NK cells, but it was not significantly different. This discrepancy could possibly be explained by the use of multiple markers to identify the NK cell by flow cytometry, and the very low number of NK cells in a lymph node. The CD56⁺ NK cells in the IL-12-treated lymph nodes produced considerable amounts of IFN-γ. The higher number of intratumoral NK cells and the high IFN-γ production by NK cells in the lymph nodes implies a prominent role of the NK cell in antitumor immunity after i.t. IL-12 treatment in HNSCC patients. A high number of CD56⁺ NK cells was correlated with a better survival, unrelated with IL-12-treatment. However, after IL-12 treatment more patients had a higher number of CD56⁺ cells than control patients. Due to the limited number of IL-12-treated patients, separation in a high and low number of CD56⁺ cells within the IL-12-treated and control patients was not possible.

The number of peritumoral CD20⁺ B cells was extremely high in some patients, but the difference between the groups was not significant. Remarkably, no differences were seen in CD8⁺ or CD4⁺ T cells. This is in contrast with other studies of rhIL-12 treatment in humans, in which histopathologic examination was done. In melanoma and cutaneous T-cell lymphoma patients, an increase in CD8⁺ cells was detected in the neoplastic lesions after s.c. IL-12 treatment; however, no B-cell markers were used in either study (29, 30). In a phase I study with s.c. IL-12 and rituximab in B-cell non-Hodgkin lymphoma, T-cell and B-cell markers were studied in sections of biopsies and a slight increase in T-cell (and not B-cell) infiltration was noticed (31). Unfortunately, these biopsies did not allow proper examination of the localization of the B cells intratumorally or peritumorally. Our results in part agree with mice studies in which effector cells required for IL-12 induced antitumor effect include NK, NKT, CD8⁺, and/or CD4⁺ cells. In these mouse models, however, the contribution of the different effector cells to the antitumor effects are mouse model and IL-12 treatment schedule dependent (32). The role of the B cell in the antitumor effect of IL-12 has not yet been described. A limited number of studies report on the role of the B cell in immunosurveillance of cancer. Both in mice and in human breast cancer there are indications of involvement of B-cell responses against tumors (33, 34).

The histologic and immunohistochemical data could not be related to toxicity, because the patients who developed severe toxicity received only one injection. Therefore, in the patients with severe toxicity the time between the single rhIL-12 injection and the surgery was much longer (8-15 days) than the planned time period (24 hours) in the patients without severe toxicity. Thus, due to this longer time period after severe toxicity, the possible immunohistochemical alterations could not be observed.

The increased size of lymph nodes resulted in a higher number of lymph nodes found in the neck dissection specimens in the IL-12-treated patients. This agrees with studies in mice (35, 36). The change in architecture of the lymph nodes after IL-12 treatment was remarkable. Fewer secondary and primary follicles were seen, although the lymph nodes were larger and the number of CD20⁺ cells was certainly not decreased. The germinal centers in the secondary follicles were smaller and CD20⁺ cells had a different distribution. To our knowledge, this is the first report describing a change in the architecture of the lymph nodes after IL-12 treatment. The organization of the lymph node and the germinal center formation is a complex process that includes B-cell proliferation and differentiation, a follicular dendritic cell network, antigen-specific T-helper cells, and tangible body macrophages. A wide variety of cytokines and chemokines influence the germinal center cell viability, proliferation, and differentiation, including immunoglobulin class switching and terminal differentiation (37). Notably, IL-12 i.t. in patients with HNSCC has effects on the B cell, both in the lymph nodes and peritumorally. The cause of the change in the architecture in the lymph nodes is currently unknown. Further studies are required to determine the causes of the change in architecture.

The change in lymph node architecture in the IL-12-treated patients is partly caused by less paracortical hyperplasia and correlates with lower numbers of DC-LAMP⁺ cells in the paracortex compared with the control patients. DC-LAMP is a lysosomal glycoprotein expressed by interdigitating DCs in human (38). DC-LAMP is localized in the MHC class II molecules-containing compartments and has been reported to be predominantly expressed by mature DCs. The finding that large amounts of DC-LAMP⁺ cells are present in the control patients and decreases in IL-12-treated patients is counterintuitive as one would rather expect enhanced DC maturation upon immune activation. However, recently it was shown that semimature DCs can induce tolerance (39, 40). The phenotypic characterization of these tolerizing DCs is not yet fully known. Therefore, it is tempting to speculate that the high number of DC-LAMP⁺ cells in the immunosuppressed control HNSCC patients is a manifestation of tolerance rather than immunity, which is reversed upon IL-12 treatment.

The number of neutrophils in the lymph nodes increased after IL-12 treatment. In the primary tumor, however, the neutrophils are slightly diminished by number in the tumor nests and remained the same peritumorally and intrastromally. There are conflicting data in animal and human studies about the role of neutrophils in cancer (41). Their presence may be detrimental by favoring malignant growth and progression. Nevertheless, recent studies have suggested that they are active in immunosurveillance against several tumors (41).

The number of mast cells in the lymph nodes of IL-12-treated patients decreased, but no differences were detected in the primary tumor. In oral squamous cell carcinoma, the number of mast cells is correlated with the number of microvessels and with a bad prognosis (42). The number of eosinophils was slightly lower in the primary tumor of the IL-12-treated patients. This agrees with studies in mice treated with IL-12 (43). Because both mast cells and eosinophils are seen as cells involved in the T helper 2 pathway, these findings were as expected considering the T helper 1 activity of IL-12 (44).

Altogether, the histopathologic and immunohistologic analysis of the primary tumors and lymph nodes of i.t. given rhIL-12 in patients with HNSCC, the most striking effect was seen on the number and IFN-γ production by NK cells, in the primary tumor and the lymph nodes, respectively. In addition, a change in the architecture of the lymph nodes was detected because of a distinct distribution of the B cells, the smaller germinal centers, and the smaller number of DC-LAMP⁺ cells. The peritumoral B-cell infiltration was very high in some IL-12-treated patients. Further investigations are required to examine the events that occur in the lymph nodes after IL-12 treatment, in particular the effects on the B cell.