Carcinoembryonic antigen-targeted photodynamic therapy in colorectal cancer models

Surgery is a cornerstone in curative treatment of colorectal cancer; however, incomplete resection can drastically reduce survival after surgery [1, 2]. Radical resection of tumors might be hampered by involvement of critical structures including large vessels, nerves, or visceral organs. In advanced-stage cancers, adjuvant local or systemic therapies can improve survival after (incomplete) resection. Survival of systemically treated patients with advanced colorectal cancer has improved with modern systemic therapies [3], but the maximum dose of systemic therapy is limited by toxicity and side effects [4].

Photodynamic therapy (PDT) could play a role in overcoming the limitations of incomplete radical resection and toxicity from systemic therapy. PDT has gained a role in treatment in different fields of oncology nowadays [5‐8]. It can be applied as standalone treatment modality, although it might also serve as adjuvant treatment to surgery after incomplete resection [9].

The principle of PDT is based on combining three non-toxic components: oxygen, light, and a photosensitizer. The photosensitizer is excited with the physical energy of non-ionizing light (of a specific wavelength) which, through a series of photochemical reactions, results in formation of highly reactive oxygen species (ROS) [10, 11]. In turn, ROS may induce local cell apoptosis and necrosis and/or cause microvascular damage. Furthermore, a change in photosensitizer structure and hydrophilicity has been proposed to contribute to cell damage [12]. Preferably, accumulation of the photosensitizer should be tumor-specific to prevent extensive damage to the normal tissue surrounding the tumor and to increase the intratumoral dose. Therefore, a tumor-targeted PDT approach has been developed [13]. In targeted photodynamic therapy (tPDT), a tissue of interest is selectively localized using a targeting vehicle conjugated to a photosensitizer. When the photosensitizer-vehicle conjugate has accumulated in targeted tissue, light of a specific wavelength is administered locally, making this therapy highly specific.

Carcinoembryonic antigen (CEA) is a membrane-anchored glycoprotein and is overexpressed in 90–95% of colorectal cancer cases. Therefore, CEA can be used for primary targeting of colorectal carcinomas. hMN-14 (labetuzumab) is an IgG directed against the carcinoembryonic antigen-related cell adhesion molecule 5 with high affinity [14]. However, different tumors express different amounts of CEA. Therefore, in clinical practice CEA-targeted PDT might only be useful in tumors with sufficient CEA expression, when complete tumor resection is hampered by the presence of critical structures that are to be preserved.

Here, we investigate whether the effect of tumor-targeted PDT is influenced by the availability of CEA on the cell surface of tumor cells with different CEA expression levels. Our multimodal conjugate (DTPA-hMN-14-700DX) consists of the humanized anti-CEA antibody, hMN-14 (labetuzumab), the photosensitizer IRDye700DX, and the chelator diethylenetriaminepentaacetic acid (DTPA). DTPA allows radiolabeling with ¹¹¹In and subsequent in vivo tumor detection and precise quantification of the antibody conjugate accumulation. After the in vitro experiments on different colorectal cancer cell lines, the therapeutic effect of tPDT was evaluated in vivo in a xenograft mouse model.

First, we assessed the binding capacity of ¹¹¹In-DTPA-hMN-14-IRdye700DX to cell lines with different CEA expression levels. This capacity can be considered as a surrogate value for accessible CEA epitopes on the cell surface. Additional file 2: Figure S1 illustrates that the cells used in this study can be roughly classified into two groups: low and high CEA accessibility.

To gain insight into the effects of tPDT, we treated 2 cell lines with low CEA availability and 2 cell lines with high CEA availability under different conditions, varying both the antibody conjugate concentration and radiant exposure (light dose).

LS174T and LoVo cells (both high CEA availability) showed a light dose-dependent increase of therapeutic efficacy of tPDT, regardless of antibody conjugate concentration (Fig. 1). At the highest light dose (314.5 J/cm²), the effect of tPDT on LoVo was higher at 3 μg/mL compared to 1 μg/mL antibody conjugate concentration (27.8% vs 15.1% viable cells, p = 0.002). At lower light doses, we did not observe an additional effect of increasing the antibody conjugate dose. Similar to LoVo, in LS174T the effect of tPDT increased with increasing light dose (Fig. 1). So, in the cell lines with higher CEA availability, we observed a clear effect of increasing light dose and antibody conjugate dose on the effect of tPDT. In contrast, in cell lines with lower CEA availability (DLD-1 and HT-29), we did not observe this variation in susceptibility for tPDT. Decreased cell viability due to tPDT in DLD-1 was only observed at the highest dose level of light (p < 0.001) and was not dependent on the antibody conjugate concentration used. In HT-29 cells (low to moderate CEA availability), we did not find an effect of tPDT at 1 μg/mL of antibody conjugate. At 3 μg/mL, however, there was an effect of tPDT at the two highest light doses (94.4 and 314.5 J/cm²). From these experiments, we concluded that the effect of antibody conjugate dose and light dose on the efficacy of tPDT seem to be dependent on the CEA availability in tumor cells.

Subsequently, we performed a separate experiment to investigate the relationship between CEA availability and tPDT effects. To explore this relationship, we included 9 cell lines with varying CEA expression levels. Figure 2 summarizes our findings on tPDT effect and its relation to amount of antibody conjugate binding of the different cell lines. No treatment effect was observed in cells with less than 5% specific binding of the antibody conjugate, which represents low CEA availability (HT-29, SW620, SW480, WiDr, and DLD-1). SW-1222 (5.5% specific binding) showed a moderate effect of tPDT after treatment (78.2% viable cells). LS174T, LoVo, and SW48 (7.7, 11.7, and 30.6% specific binding, respectively) were more sensitive to treatment with tPDT (47.2%, 50.1%, and 48.1% viable cells after treatment, respectively), suggesting that a minimum level of antibody conjugate binding to the cells is required for effective treatment with tPDT.

Subsequently, we evaluated the ability of tPDT to reduce in vivo tumor progression in a xenograft mouse model. Before evaluating the treatment, we performed an ex vivo biodistribution of ¹¹¹In-labeled DTPA-hMN-14-IRDye700DX. We found 13.7 ± 3.8 %ID/g of our tracer in tumor, 8.8 ± 0.5 %ID/g in spleen and 29.4 ± 2.6 %ID/g in liver tissue (Additional file 2: Figure S1). In mice treated with tPDT, we generally observed slower tumor growth than the control groups (n = 17, p > 0.05). The tumor of one mouse in the treatment group, however, showed a more aggressive growth pattern after treatment (mouse 1, Fig. 3). This was possibly caused by incomplete tumor irradiation during tPDT, but because this was not reported during treatment of this mouse, this dataset was not rejected from the analyses.

The current study demonstrates that in vitro tumor cell lines with different levels of CEA targetability have varying susceptibility to tPDT. Furthermore, the amount of ¹¹¹In-DTPA-hMN-14-IRDye700DX binding to these cells seems to play a crucial role in the efficacy of tPDT. Our findings suggest that only cell lines that can be targeted with higher amounts of antibody-conjugate can be treated effectively with tPDT. In vivo, we observed a trend towards tumor growth delay in xenografted tumors of tPDT-treated mice.

In the tested cell lines, higher light dose increased the efficacy of tPDT. However, in DLD-1 and LS174T cells, the maximum light dose led to a statistically significant decrease in cell viability even without the presence of the antibody conjugate (Fig. 1). This suggests that the energy provided by the light source may have been excessive in this condition. Hyperthermia can lead to decreased cell survival in (monolayer) cells and is likely to be responsible for the decrease in cell viability of the 0 μg/mL control condition of LS174T and DLD-1 [16, 17].

In this study, we included a large series of CEA-expressing cell lines in a direct comparison. To compare the tPDT effect in all cell lines under identical conditions, we performed a single experiment with fixed conditions: 0.5 μg/μL ¹¹¹In-DTPA-hMN-14-IRDye700DX and total radiant exposure 94.5 J/cm². Higher radiant exposures could have an additional thermal effect on several cell lines (see above). Since we chose 94.5 J/cm², the observed effects depicted in Fig. 2 are considered to be only tPDT-mediated, but may be an underestimation of the maximum tPDT effect that could be achieved. More specific binding of ¹¹¹In-DTPA-hMN-14-IRDye700DX resulted in a larger treatment efficacy of tPDT in vitro (Fig. 2). These findings are consistent with previous studies that showed an increased efficacy of tPDT when there is more photosensitizer available in the tumor [13].

Increasing the number of photosensitizer moieties per antibody molecule (substitution ratio) might further increase the amount of photosensitizer in tumors. However, as we use a random conjugation method, excessive conjugation of the photosensitizer to an antibody might affect the antibody binding affinity. Additionally, when conjugating high amounts of a photosensitizer or in general a fluorescent dye to an antibody, changes in chemical properties (lipophilicity and net charge) may lead to faster blood clearance, less tumor uptake, and more accumulation in liver and spleen [18‐20]. Findings from these studies are in line with the findings of the biodistribution performed in the current study. Random conjugation strategies can, therefore, be considered a limitation regarding the preparation of an antibody-photosensitizer conjugate. Our results also suggest that tPDT efficacy is dependent on a combination of radiant exposure and amount of photosensitizer that is present on the target. In addition, different coping strategies of tumor cells to tPDT-induced damage may affect treatment efficacy, but this was not taken into account in the current study.

In contrast to in vitro tPDT, in vivo tPDT can be particularly challenging due to the multifactorial nature of the treatment and treatment effect. In the current study, we tested the efficacy of tPDT with subcutaneous LoVo tumors in BALB/c nude mice. LoVo cells were chosen as a representative because the antibody-conjugate binding to the cells was the most representative of the 3 cell lines with the a highest binding percentage (SW-48, LoVo, and LS174T) (Additional file 2: Figure S1). We observed a steady increase in tumor size in the control groups during 6 weeks. The treated mice showed a slower tumor growth pattern in general (Fig. 3). The delay in tumor growth was, however, not statistically different compared to the control groups. This was caused by one treated mouse that showed an accelerated tumor growth pattern. When cells survive after (incomplete) tPDT, cell cycle progression might be stimulated, resulting in increased proliferation, and invasive and metastatic growth [21]. This can particularly occur when treatment is not adequate. Furthermore, in a single tumor, heterogeneity in CEA expression may exist, which could be a limitation in adequate tumor targeting [22, 23]. Earlier studies indicate that host immunity can also play a crucial role in achieving a successful effect of (targeted) PDT [10, 24].

However, due to the use of an immunodeficient mouse model, the effects of immune system involvement could not be assessed in the current study. Moreover, fractionated and repeated tPDT might enhance the tPDT effect, as observed earlier by Mitsunaga and colleagues [25].

During the past years, targeted PDT approaches have gained more clinical and scientific interest. In the clinical situation, tPDT may mainly have a role as adjuvant therapy to surgery [9]. The main advantage of tPDT over other adjuvant therapies is the ability to selectively apply therapy to local areas and tissues of interest, thereby minimizing side effects. When combined with NIR fluorescence or nuclear imaging, its potential for clinical use as theranostic approach is highlighted even further [26, 27].

Here, we focused on tPDT in a colorectal cancer model using a humanized monoclonal antibody directed against CEA (CEACAM5). Since other primary epithelial malignancies also express varying levels of CEA, CEA-targeted PDT could potentially be used in treatment of other malignancies as well. These include carcinomas of the gall bladder, urinary bladder, stomach, pancreas, ovary, endometrium, and lung [28]. Other conjugates of the same antibody have already been used in (clinical) studies investigating fluorescence and radioguided surgery, radioimmunotherapy, and antibody-drug conjugates [29‐35].

Our findings are in line with work by Shirasu et al., who showed phototoxic and dose-dependent effects of CEA-targeted PDT, with their anti-CEA antibody C2-45 conjugated with IRDye700DX [36]. Our radiolabeled antibody conjugate, ¹¹¹In-DTPA-hMN-14-IRDye700DX, allows precise quantification of the tracer both in vivo and in vitro, and may enable SPECT or PET imaging prior to surgery. During surgery, when tumors are covered by overlying tissue, the radiolabel could guide the surgeon to the area of interest, using a gamma probe prior to resection of tumors [29, 37]. Subsequently, tumor margins may be visualized using near-infrared fluorescence imaging using the same radiolabeled antibody conjugate. Lastly, the photosensitizer can be irradiated to treat residual tumor that was excluded from excision [9].

Our findings suggest cells (and tumors) have different levels of susceptibility for tPDT even though they all express CEA. Furthermore, tPDT can effectively reduce tumor growth in vivo. In the current study, we treated whole subcutaneous xenografted tumors, whereas in the clinical situation one would optimally use tPDT after resection to treat (microscopically) small amounts of residual tumor cells. Thus, our findings could even be an underestimation of clinically achievable results with tPDT.