Precision Medicine for Gastric Cancer: Current State of Organoid Drug Testing

For review article

3. Response to Comments on the Quality of English Language

Comment: English language was well-written and no visible errors were detected. There were some inconsistencies in spacing, punctuation, and formatting, which can be easily improved with minor editing.

Response 1:    Thank you for your comment. We have now made some changes to be consistent with our spacing, especially by adding a space before the brackets of a reference, and some minor editing in formatting.

Response to Reviewer 2 Comments

1. Summary

Thank you very much for taking the time to review this manuscript. We truly appreciate your insightful feedback, which has helped clarify and strengthen our arguments and significantly improved the overall quality of our work. Below, you will find our detailed responses to each comment in blue, along with the corresponding corrections highlighted in red in the relevant paragraphs of the re-submitted file.

2. Point-by-point response to Comments and Suggestions for Authors

The authors provide a comprehensive review on the use of gastric cancer (GC) organoids in precision medicine, filling a clear gap in the literature by focussing on the correlation between drug responses in patient-derived GC organoids and clinical outcomes. The inclusion of summary tables and synthesis of findings from multiple studies adds clarity and depth to the review. Additionally, the authors highlight the variability in experimental methods and suggest considerations for design improvements. However, a more explicit recommendation on which or how methodologies should be standardised could be helpful. Overall, the review was well-structured, clearly written and provided a balanced view of the field, integrating findings from key papers.

Section 1: Introduction

Comments 1: Consider including a sentence or two about the heterogeneity of gastric cancer (e.g. the different molecular/ histologic subtypes) in paragraph 1. This would provide better context for the paper.

Response 1: Thank you for pointing this out. To address this comment, we have now added a few sentences to explain the heterogeneity of gastric cancer in paragraph 1 (lines 30-37).

“GC is a heterogeneous disease that is mostly composed of gastric adenocarcinoma, that is further sub-classified into intestinal, diffuse, unclassified, or intermediate types[5], with significant molecular heterogeneity identified by The Cancer Genome Atlas (TCGA) to include EBV-positive (EBV+), microsatellite instable (MSI), genomically stable, and chromosomally instable (CIN) subtypes[6]. This heterogeneity of GC significantly impacts survival outcomes. Exploring GC treatment options that can effectively mitigate the impact of GC is crucial for this poor prognosis cohort of patients.”

Comments 2: Lines 70-75: Is this statement correct? There have been several papers that have attempted to link genomic profiles to chemotherapeutic outcomes for patient stratification, but perhaps not to the extent of functional assays. E.g. there’s an ARID1A study (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10110689/) but it was conducted retrospectively. Also later in the review, section 3.5 discusses molecularly targeted therapies, where it is suggested that organoids with certain mutations are more sensitive to some drugs

Response 2: We appreciate the reviewer’s insightful comment and thank you for bringing this to our attention. We have removed the previous statement and included the ARID1A mutation alongside the Dihydropyrimidine Dehydrogenase (DPYD) variant to emphasize that these markers can aid in predicting treatment outcomes. Additionally, we highlight how we can integrate these genomic profiles with a functional assay to better inform personalized treatment strategies (lines 74-78).

“Next-generation sequencing (NGS) technologies enable a thorough analysis of tumour (epi)genomics and transcriptomics, providing an in-depth understanding of the molecular characteristics specific to each tumour [20] including TMB, a key indicator of potential sensitivity to ICI [21]. These developments have paved the way for careful categorisation and detection of germline or tumour-specific markers. Examples include, Dihydropyrimidine Dehydrogenase (DPYD) variants found in 3-5% of Caucasian, African American and Asian populations that predict poor response to 5-FU [22], and ARID1A mutations that predict favorable overall survival outcomes to fluorouracil based chemotherapeutics and pembrolizumab for PD-1 blockade[23]. Even though potential sensitivity to chemotherapeutics and targeted therapies can be identified by tumour sequencing, clinical outcomes from cancer treatment led purely by this static measure of genomic data have been largely inadequate [24]. Therefore, integrating tumour genomic profiles with functional assays, such as in-vitro models, may enhance precision medicine efforts to guide personalized treatment strategies for GC.”

Comments 3: Line 105-107: The Ren et al. review includes gastric cancer too and this should be added.

Response 3: Thank you for this helpful comment. We have revised the wording from “stomach” to “gastric” to enhance clarity and avoid any potential confusion (line 115).

“A detailed review of drug-testing methodologies used in GC organoid studies is lacking. Ren et al. provided an overview of the basic technology and clinical applications of drug screening using organoids across various cancers including colorectal, liver, gastric, pancreatic, and brain cancers  [45].”

Section 2: Gastric Cancer Patient-Derived Organoid Establishment

Comments 4: Line 132-133: is it not concerning that majority of their organoid lines do not survive past P10, and the paper also indicates different organoid morphology. Could the authors comment further on this..?

Response 4: Thank you for bringing this to our attention. In addition to the difference in genomic expression in low and high-growth rate organoids that the study suggested, we have now added several sentences to highlight that the majority of the organoids (91%) did not survive past passage 10 (P10) and provided some insights into the possible reasons for this in lines 146-149.

“However, given the majority of organoid lines (91%) in this study did not survive past passage 10, it suggests that there may be essential components missing in the growth media to maintain long term growth, or that some of the organoid lines may be contaminated by normal gastric organoids, as can be indicated by cystic morphology.”

Comments 5: Lines 156-173: the authors raise an important challenge regarding GC organoid culture, however, they stop short of specific recommendations on how to enrich for GC organoids. Should researchers consider testing alternative growth media going forward? This suggestion can be made more explicit

Response 5: We appreciate the reviewer’s comment here. As a result, we have included a couple of sentences to emphasize that a combination of two methods (alternative media and manual selection) should be employed moving forward to ensure a pure gastric cancer organoid population. These changes can be found in lines 177 to 180.

“This methodology from Nanki et al. provides the current, best-practise path to GC organoid culture success, although combining alternative medias with manual selection of GC organoids away from normal gastric organoids would further elevate the technique to ensure a pure population of GC organoids is established.”

Comments 1: Line 174-175: how would one verify that the cultured cells are tumorigenic? It would be helpful to suggest specific assays to assess tumorigenicity.

Response 1: Thank you for highlighting this important point. We appreciate your feedback. We have now proposed some methods to assess tumorigenicity, which can be found in lines 190-193.

“This can be performed using the fairly time-consuming process of transplant into immunocompromised host mice and detection of tumour formation or more rapidly by validation of the presence of tumour-associated alterations via genomic sequencing.”

Section 3: Gastric Cancer Organoid Drug Testing: Summary of Current Methodologies

Comments 2: To consider incorporating new paper into Section 3 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11323579/

Response 2: Thank you for the suggestion. We have now incorporated the paper into Section 3, specifically in line 319 where we included the study's reference. Additionally, we have added details about this study in the last rows of Tables 1 and 2.

“3.5. Testing Molecularly Targeted Therapies

A handful of studies tested the sensitivity of GC organoids to molecularly targeted drugs, ranging from the evaluation of 1-3 drugs in clinical use for GC to investigational panels of over 30 targeted drugs [53, 62, 63]”

Comments 3: Table 1 is useful but I wondered if it could be improved by including information on the starting material used in each paper. Perhaps the location of tissue collected as well as tumour subtype?

a.      Consider removing “total no. of GC patients recruited”, because “efficiency of GC organoid establishment” appears to be calculated using “total no. of GC organoids established” as a percentage of “total no. of GC patients recruited”.

Response 3: We appreciate your suggestion. We have removed the column titled “Total GC Patients Recruited” from Table 1 and added three additional columns: “Tissue Acquisition Method” to include starting material, “Tumor Site,” and “Histology (Lauren Classification)” to specify the subtype of gastric cancer.

Comments 4: Line 205-207: If the authors believe that assessing culture quality is important for future studies, they should explicitly recommend this, and clarify which metrics should be used.

Response 4: Thank you for pointing this out. We agree with this comment and have now added our suggestions for future studies to assess culture quality, ensuring that they accurately represent the patient tumour. These changes can be found in lines 229-235.

“We suggest that future studies use PDOs with sufficient expansion capacity for drug testing, with the amount determined by the number of drugs being tested and layout/size of the screening setup, and that genetic sequencing be utilised to assess the tumorigenicity of all PDOs and to confirm that they accurately represent the original patient tumour. This validation is crucial for ensuring that organoid responses model tumour response and can effectively inform patient outcomes.”

Comments 5: Section 3.2: It is unclear when the optimal time is to initiate drug treatment for the organoids, as well as the recommended duration of treatment. Clarification on this would be beneficial.

Response 5: Thank you for your feedback. We’ve added a few sentences at the end of section 3.2 to emphasize that, although the field still requires larger studies to standardize these methods, we can adapt the methodologies from gastric organoid studies that have effectively matched organoid drug responses to patient clinical outcomes. Please see the revisions in lines 249-253 below.

“While the field awaits larger studies reporting GC organoid drug responses and clinical outcomes to standardize these procedures, future studies could aim to initiate drug treatment 2-3 days post seeding and continue treatments for 4-6 days. These protocols align with the GC organoid studies that have validated PDO drug responses against clinical outcomes thus far.”

Comments 6: Section 3.4: the authors comment on how different papers use different methodologies to test both single drugs and drugs in combination, as well as highlight some issues with the various methods (e.g. limited correlation to clinical response). However, in lines 280-282 the authors conclude by saying any of the methodologies may be adequate. It would more helpful to suggest a specific methodology that is more widely recognized or demonstrates better correlation with clinical responses.

Response 6: Thanks for your comment here. We have removed the phrase stating that “any of the drug treatment methodologies may be adequate to predict patient responses” and added a few sentences at the end of section 3.4. These additions highlight specific methods that could enhance drug testing moving forward. Please refer to lines 306-315 for these updates.

“Overall, GC organoid responses to combination drug treatment in vitro across these studies were similar to patient responses. However, only a very small number of patients with known clinical response data were included, thus confirmation of the predictive utility of GC organoid drug testing in the clinic awaits further testing in larger patient cohorts. Concurrently, it is essential to standardise methods when treating PDOs with chemotherapeutics. Utilizing a concentration range that spans therapeutic dosing is crucial for both mono-chemotherapeutics and combinations. Additionally, when conducting combination chemotherapy regimens on PDOs, agents should be added together instead of evaluating the responses of these agents individually as this requires a smaller amount of GC organoid starting material to speed up the process [65]. “

Section 4: Methods of Comparing Organoid Drug Response to Patient Response in the Clinic

Comments 7: Lines 320-322: define huTGO4. The authors mention this paper where organoid response to 5FU, oxaliplatin and epirubicin corresponded to patient clinical treatment response in some human tumour organoid lines but not others. However it is unclear why this is the case— it would be helpful to include a brief discussion on potential reasons for this discrepancy.

Response 7: We thank the reviewer for pointing this out. We have removed the term “huTGO4” and now refer to it as the “other PDO” to clarify that it denotes the second organoid line discussed in the study. Additionally, we have included an explanation of the discrepancy between the organoid and patient responses in lines 350-355.

“In contrast, the other PDO was relatively sensitive to treatment, which is inconsistent with the patient's lack of response to chemotherapy [29]. This discrepancy was explained by the lack of immune component in PDOs, and that the patient may have high levels of infiltrating myeloid suppressor cells that suppress T cell activation and can lead to poor tumour response. In the future this may be addressed with the inclusion of immune cell populations in coculture with  PDOs [68].”

Comments 8: Figure 2: the authors should edit the bottom panel to include the total number of patients assessed too and not just the number with matching responses as this would provide clearer context.

Response 8: Thank you for your suggestion, which provides a clearer picture of the comparisons between organoid and patient responses. We have adjusted the bottom panel of Figure 2 to include both the total comparisons made and the matched comparisons in each study, illustrating the efficacy of organoids in predicting patient tumour responses.

Comments 9: Lines 340-367: Nice informative description of Zhao et al. and Schmache et al. studies but the authors might consider shortening them slightly for conciseness.

 Response 9: We appreciate the reviewer’s comment on this matter. We have made efforts to shorten this section; however, some sentences are challenging to condense due to the complexity of the methods used to categorize patient and organoid responses to drugs (Lines 374-401).

Section 6: Conclusions

(this comment is with respect to either section 4 or conclusions)

Comments 10: An open question in this field is the feasibility of generating and characterizing cancer organoids from a patient’s biopsy within the timeframe before start of treatment. Could the authors provide insights on the current state of the field regarding this issue? Additionally, it would be beneficial to reference relevant articles that discuss high-throughput drug screening conducted in organoid models, such as https://academic.oup.com/jmcb/article/12/8/630/5873160?login=false

Response 10: Thank you for highlighting this important point. We appreciate your suggestion and have added a separate paragraph addressing the timeliness of organoid generation and the production of drug results in the current gastric cancer organoid literature. Additionally, we included a few sentences in the same paragraph to highlight other studies that have incorporated high-throughput drug screening using organoids. Please refer to lines 409-419 for these updates.

“An important aspect to explore is also the timeliness of drug result production using PDOs, as this is crucial to guiding individualised treatments. While initial studies did not specify the time required to expand GC PDOs and deliver drug results, more recent research has started to offer some insights. From the few recent studies, organoid drug screening was achieved within 2-3 weeks, however, it remains unclear whether this timeframe includes the entire duration for testing all drugs and if it accounts for the time needed for sample processing and expansion [49, 50, 53, 57]. Studies involving other organoid models have successfully utilised fully automated robotic systems to conduct drug assays [70, 71], which could help reduce expansion time, reduce the number of organoids required in smaller plating volumes, minimize the need for Matrigel to decrease costs and variability, and enhance reproducibility.”