MVA Track 2：從一個孩子的 BUB1B 到可重定位的核准藥MVA Track 2: from one child’s BUB1B to repositioned approved drugs

TMU-IIH PharmOmics。台北醫學大學跨領域學院與藥學院。通訊作者 Hung-Lin Chen。Hugging Face 帳號 hunglinchen2003。一人代表團隊送出；評審看的是最新一筆。TMU-IIH PharmOmics. Taipei Medical University, College of Interdisciplinary Studies and College of Pharmacy. Correspondent: Hung-Lin Chen. Hugging Face account hunglinchen2003. One member submits. Judges review the latest submission.

比賽：SageBio Rare Disease, Real Kid · MVA Hackathon 2026。Track 2 要的是已上市藥物的新用途，而且要說清楚機制。Challenge: SageBio Rare Disease, Real Kid · MVA Hackathon 2026. Track 2 asks for a new use of a market-approved drug, with the mechanism stated.

Track 2 的方法就是 FDA 核准藥篩選，圖一。先確認這個孩子是 BUB1B 功能喪失，再把已上市藥對上 Lys1002，留下癌症或生長失敗相關的用途，用 DynamicBind 親和力排序，最後用 knockdown 細胞核來讀。

The Track 2 method is the FDA-approved drug screen in Figure 1. Confirm BUB1B loss of function, dock marketed drugs on Lys1002, keep cancer or growth-failure uses, rank by DynamicBind affinity, and read a knockdown nucleus.

1. 八個 HPO 詞加上 VCF（5,012,204 個變異）。只有 BUB1B 是複合異型合子：p.Leu737Ter 與 p.Asn1002Lys。機制是紡錘體檢查點的功能喪失。Eight HPO terms plus the VCF (5,012,204 variants). Only BUB1B is compound heterozygous: p.Leu737Ter and p.Asn1002Lys. The mechanism is loss of function of the spindle-assembly checkpoint.
2. HY-L066，AutoDock Vina 1.2.7，Lys1002 NZ 上的 22 Å 盒子。2,432 個配體有分數，1,339 個標成 FDA。53 個中藥單體放在同一個盒子。前 100 名的門檻是 pazopanib，−6.491 kcal/mol。HY-L066, AutoDock Vina 1.2.7, a 22 Å box on Lys1002 NZ. 2,432 ligands scored, 1,339 flagged FDA. Fifty-three herbal constituents went into the same box. The top-100 cutoff is pazopanib at −6.491 kcal/mol.
3. 留下腫瘤標靶，或生長激素、癌症惡病質的註記。排除 Aurora、PLK、微管藥、雷公藤紅素，以及研究代碼 c-Kit-IN-1。沒有中藥單體進入最後面板。Keep an oncology target, or a growth-hormone or cancer-cachexia annotation. Exclude Aurora, PLK and microtubule poisons, celastrol, and the research code c-Kit-IN-1. No herbal constituent entered the final panel.
4. 十四個已上市藥依 DynamicBind 對激酶的預測親和力排序。距離不是門檻。優先六名單藥測試：pralsetinib 7.99、lapatinib 7.48、tucatinib 7.07、entrectinib 6.89、capmatinib 6.71、ponatinib 6.70。不是複方，也不是兒童處方。核准用途還要腫瘤的體細胞突變分析。Fourteen marketed drugs are ordered by DynamicBind affinity for the kinase. Distance is not the filter. The lead six are tested singly: pralsetinib 7.99, lapatinib 7.48, tucatinib 7.07, entrectinib 6.89, capmatinib 6.71, ponatinib 6.70. Not a combination and not a paediatric regimen. A labelled use still needs tumour somatic analysis.
5. 讀值是 HeLa BUB1B knockdown 的細胞核，不是 N1002K knock-in。細胞核 88 到 43。命中是這些數字回到對照。Knockdown 讓同一個讀值可以用在其他 BUB1B 基因型，以及其他紡錘體組成或代謝異常。The readout is a HeLa BUB1B-knockdown nucleus, not an N1002K knock-in. Nuclei fell from 88 to 43. A hit moves those numbers toward control. Knockdown keeps the same readout usable for other BUB1B genotypes and for other defects of spindle assembly or spindle metabolism.

停掉、而且不是這份方法的路：全長對接（Cα RMSD 33.5 Å）、激酶摺疊救援（空載 49.9，最好的配體 47.4）、115 個分子的生成沒有勝過種子，以及 2,427,793 個 ChEMBL／ZINC 姿態（表 5）。那些都不是 FDA 藥。Stopped, and not this method: full-length docking (Cα RMSD 33.5 Å), kinase-fold rescue (apo 49.9, best ligand 47.4), a 115-molecule generator that did not beat the seeds, and 2,427,793 ChEMBL/ZINC poses (Table 5). Those are not FDA drugs.

無效等位基因是 chr15:40209701 T>G，p.Leu737Ter，stop_gained，HIGH，深度 46，等位深度 21/25，VAF 約 54%。終止密碼在 1,050 個胺基酸的約 70% 處，拿掉 C 端激酶域，是典型的無義媒介衰變受質。The null allele is chr15:40209701 T>G, p.Leu737Ter, stop_gained, HIGH impact, depth 46, allelic depths 21 and 25, VAF about 54%. The stop is at about 70% of the 1,050-residue protein and removes the C-terminal kinase. It is a classical nonsense-mediated-decay substrate.

第二個等位基因是 chr15:40220612 T>G，p.Asn1002Lys，missense，MODERATE，深度 28，等位深度 15/13，VAF 約 46%。SIFT 0.00，PolyPhen-2 0.99，CADD 24.5。它在激酶 C 葉螺旋 αG 的 N-cap，離已發表的熱點 p.Leu1012Pro 十個殘基。ATP 位點的 Lys795 只移動 1.03 Å。1002 的 Cα 移動 4.07 Å。這比較像蛋白質不穩定，而不是催化位點壞死。新的離胺酸側鏈溶劑可及表面從 27 增到 82 Ų。它在幾何上可以被泛素化，我們沒有測泛素。The second allele is chr15:40220612 T>G, p.Asn1002Lys, missense, MODERATE, depth 28, allelic depths 15 and 13, VAF about 46%. SIFT 0.00, PolyPhen-2 0.99, CADD 24.5. It is the N-cap of helix αG, ten residues from the published hotspot p.Leu1012Pro. The ATP-site lysine, Lys795, moves only 1.03 Å. The Cα at 1002 moves 4.07 Å. That fits instability better than a dead catalytic site. Side-chain SASA rises from 27 to 82 Ų. The new amine could be ubiquitylated. We did not measure ubiquitin.

沒有親代 DNA，所以反式排列是依合子型與疾病模型推論。Track 1 這對變異的信心是 0.95。cDNA 編號沒有另寫，因為沒有 RNA。There is no parental DNA, so trans configuration is inferred from zygosity and the disease model. Track 1 confidence for the pair was 0.95. No cDNA nomenclature was invented, because there is no RNA.

送進細胞實驗的面板是表 6 的十四個已核准藥。優先做前六名。它們和這個孩子的連結是橫紋肌肉瘤所代表的腫瘤表型，不是宣稱這顆腫瘤帶有 RET、HER2 或 MET 驅動變異。Macimorelin 是非激酶的生長臂，對應矮小、體重不增，以及它在癌症惡病質文獻中的位置。Nilotinib 留下，因為它是剛性口袋裡最緊的 Vina 分數。Infigratinib 留下，因為它本來就在較早的 cover 短名單。The cell-assay panel is the fourteen approved drugs in Table 6. Run the first six first. Their link to this child is the tumour phenotype represented by rhabdomyosarcoma, not a claim that the tumour carries a RET, HER2 or MET driver. Macimorelin is the non-kinase growth arm, for short stature, failure to thrive, and its place in the cancer-cachexia literature. Nilotinib stays because it is the tightest rigid-pocket Vina score. Infigratinib stays because it was already on the earlier cover shortlist.

每一個優先藥的理由可以分開看。Pralsetinib 的 DynamicBind 親和力最高（7.99 / 7.85），而且 PLIP 看到 K1002 氫鍵。Lapatinib 與 tucatinib 是 HER2 藥；tucatinib 在六個裡 Vina 最緊（−7.718），對 K1002 有最近 2.06 Å 的氫鍵。Entrectinib 與 ponatinib 的親和力夠高，但 PLIP 沒有直接打到 K1002，所以它們是親和力候選，不是離胺酸蓋子。Capmatinib 最近重原子 3.1 Å，有一條 2.32 Å 氫鍵。Each lead drug has its own reason. Pralsetinib has the highest DynamicBind affinity (7.99 / 7.85) and a PLIP hydrogen bond to K1002. Lapatinib and tucatinib are HER2 drugs; tucatinib is the tightest Vina score of the six (−7.718) and has a 2.06 Å hydrogen bond to K1002. Entrectinib and ponatinib rank on affinity, and PLIP does not place them in direct contact with K1002, so they are affinity candidates rather than a lysine cover. Capmatinib’s nearest heavy atom is 3.1 Å, with one 2.32 Å hydrogen bond.

前六名是篩選順序，不是處方，也不是複方。2026 年 10 月對照的美國標示（Gavreto 2025 年 12 月、Rozlytrek 2024 年 1 月，以及 Tukysa、lapatinib、Tabrecta、Iclusig 至 2025 年的版本）裡，只有兩個藥寫了兒童年齡，而且都綁在這個生殖系 VCF 沒有的體細胞驅動變異。The lead six are a screening order, not a prescription and not a combination. United States labels consulted in October 2026 (Gavreto, December 2025; Rozlytrek, January 2024; Tukysa, lapatinib, Tabrecta and Iclusig through 2025) give a paediatric age to only two of them, and only with a somatic driver that this germline VCF does not contain.

Pralsetinib：成人，以及 12 歲以上、晚期或轉移性 RET 融合甲狀腺癌、需要全身治療，而且在適合用放射性碘時必須是碘治療無效。未滿 12 歲未建立，兒童的 RET 融合肺癌也未建立。標示要求監測未閉合的生長板；動物在接近臨床暴露時出現生長板發育異常。這個孩子已經有生長失敗。Entrectinib：大於 1 個月，實體瘤要有 NTRK 融合、沒有已知抗藥突變，而且已經轉移、開刀會造成嚴重損傷，或沒有其他合適治療。ROS1 肺癌只有成人。標示記載兒童骨折增加，有些沒有外傷。Lapatinib、tucatinib、capmatinib、ponatinib：兒童的安全性與有效性都未建立。Ponatinib 的幼年大鼠在 3 mg/kg/日、約成人 45 mg 的 0.32 倍（以 mg/m² 計）出現死亡；成人標示有動脈阻塞、靜脈血栓、心衰竭與肝毒性的黑框。六個藥都沒有 BUB1B 或 MVA 的適應症。這個孩子的橫紋肌肉瘤是檢查點表型，不是 RET、NTRK、HER2、MET 或 BCR-ABL 驅動的證據。Pralsetinib is labelled for adults and for patients 12 years and older with advanced or metastatic RET fusion-positive thyroid cancer that needs systemic therapy and is radioactive-iodine refractory when iodine is appropriate. It is not established under 12 years, or for paediatric RET fusion-positive lung cancer. The label says to monitor open growth plates; animals showed physeal dysplasia near the clinical exposure. This child already has growth failure. Entrectinib is labelled for patients older than 1 month with an NTRK fusion-positive solid tumour that is metastatic or whose resection would cause severe morbidity, that has progressed or has no satisfactory alternative, and that lacks a known acquired resistance mutation. ROS1 lung cancer is adult. The label reports fractures in children, including fractures without trauma. Lapatinib, tucatinib, capmatinib and ponatinib: paediatric safety and effectiveness have not been established. Ponatinib’s juvenile-rat study recorded deaths at 3 mg/kg/day, about 0.32 times the 45 mg human dose on a mg/m² basis, and the adult label has a boxed warning for arterial occlusion, venous thromboembolism, heart failure and hepatotoxicity. None of the six is labelled for BUB1B or for MVA. This child’s rhabdomyosarcoma is the checkpoint phenotype, not evidence of a RET, NTRK, HER2, MET or BCR-ABL driver.

若將來要考慮這些藥的核准用途，腫瘤本身要先做體細胞突變分析，而且要跟藥對得上：pralsetinib 要 RET 融合，entrectinib 要 NTRK 融合，lapatinib 或 tucatinib 要 HER2，capmatinib 要 MET exon 14 skipping，ponatinib 要 BCR-ABL1。生殖系 BUB1B 不能代替這項檢驗。本研究沒有定序這顆腫瘤。Any later look at a labelled use would first require somatic mutation analysis of the tumour, matched to the drug: a RET fusion for pralsetinib, an NTRK fusion for entrectinib, HER2 for lapatinib or tucatinib, MET exon 14 skipping for capmatinib, and BCR-ABL1 for ponatinib. Germline BUB1B does not replace that test. This study did not sequence the tumour.

六個藥不是複方。標示沒有把彼此寫成具名禁忌，也沒有核准六個一起用。已核准的合併是一個激酶抑制劑加上抗體或化療。Lapatinib 與 tucatinib 都打 HER2，臨床報告是先後使用，不是同時用。Tucatinib 是強 CYP3A 抑制劑，標示要求避開「濃度稍升就會出嚴重毒性」的 CYP3A 受質；另外五個都是 CYP3A 受質。Pralsetinib 另要求避開中至強 CYP3A 抑制劑與 P-gp 抑制劑，而 lapatinib 會抑制 CYP3A4 與 P-gp。肝、肺與 QT 的警告會疊加。Ponatinib 的血管與肝臟黑框使它最不適合放進任何合併。Track 2 允許已上市藥合併，但每個藥要有自己的理由。這裡六個是同一個細胞核讀值的替代候選，實驗一次測一個。只有兩個單藥各自把 knockdown 細胞核拉回對照，才在細胞實驗裡做一對。那仍然不是固定複方，也不是兒童劑量。The six are not a combination. Labels do not name one as a contraindication to the others, and they do not approve the six together. Approved combinations pair one kinase inhibitor with an antibody or with chemotherapy. Lapatinib and tucatinib both inhibit HER2; clinical reports use them in sequence, not together. Tucatinib is a strong CYP3A inhibitor, and its label says to avoid CYP3A substrates when a small concentration rise can be serious. The other five are CYP3A substrates. Pralsetinib’s label also says to avoid moderate or strong CYP3A inhibitors and P-gp inhibitors, and lapatinib inhibits CYP3A4 and P-gp. Liver, lung and QT warnings would stack. Ponatinib’s vascular and hepatic boxed warning makes it the member least suited to any combination. Track 2 allows marketed drugs to be combined when each has its own reason. Here the six are alternative candidates for one nucleus, so the assay tests them singly. A pair belongs in the cell experiment only after both singles move the knockdown nucleus toward control. That pair would still not be a fixed-dose product or a paediatric dose.

臨床相容性與毒性排除Clinical compatibility and toxicity exclusion

表 6 是親和力順序。下面這一層用的是病歷裡寫下的表型：骨骼肌萎縮、腎鈣質沉著、橫紋肌肉瘤、32 週早產、出生體重約 1 公斤、身材矮小、體重不增。八個 HPO 沒有神經系統發育遲緩。中樞神經暴露只在標示本來就為它設計時才列入。這一層不是劑量，也不能代替腫瘤體細胞定序。Table 6 is the affinity order. The tier below uses the recorded phenotype: skeletal muscle atrophy, nephrocalcinosis, rhabdomyosarcoma, birth at 32 weeks, birth weight about 1 kg, short stature and failure to thrive. The eight HPO terms do not include developmental delay. Central-nervous-system exposure is weighed only where a label was built for it. This tier is not a dose and does not replace tumour somatic genotyping.

Entrectinib 是為進入中樞神經而開發的。標示對大於 1 個月、帶 NTRK 融合的實體瘤，按年齡與體表面積給劑量。肉瘤只有在有 NTRK 融合時才落在這個適應症裡。這個孩子的橫紋肌肉瘤還沒有定序，所以對上的是標示上的腫瘤類別，不是宣稱這顆瘤是 NTRK 陽性。標示已要求監測骨折與認知影響。Pralsetinib 在這一梯隊裡選擇性較高。它的實際優點是單藥、而且不要併用中至強 CYP3A 或 P-gp 抑制劑。Azole 類抗黴菌藥就在要避開的那一組。未滿 12 歲的安全性未建立，未閉合的生長板仍要監測。Entrectinib was developed for central-nervous-system exposure. The label gives doses by age and body surface area for patients older than 1 month with an NTRK-fusion solid tumour. A sarcoma is inside that indication only when an NTRK fusion is present. This child’s rhabdomyosarcoma has not been sequenced, so the match is to the labelled tumour class, not a claim that the tumour is NTRK-positive. Fractures and cognitive effects are labelled risks. Pralsetinib is the more selective name in this tier. Its advantage is single-agent use without a moderate or strong CYP3A or P-gp inhibitor. Azole antifungals are in that group. Safety under 12 years is not established, and open growth plates must be monitored.

Tucatinib 在 HER2 陽性腦轉移有中樞活性，而且比同時抑制 EGFR 的 lapatinib 更選擇 HER2，皮疹與腹瀉因此不是它的主要問題。它是強 CYP3A 抑制劑，一些止吐藥、鎮靜藥和抗感染藥若走 CYP3A，就很難跟它放在一起。Capmatinib 留下，是因為 MET 訊息參與骨骼肌生物，也在橫紋肌肉瘤被研究過。那是生物關聯，不是這個孩子的核准用途。標示有間質性肺炎警告。對 32 週出生的孩子，這個肺部警告使它停在第二梯隊。Tucatinib has central-nervous-system activity in HER2-positive brain metastases and is more HER2-selective than lapatinib, so rash and diarrhoea are less dominant than with an EGFR/HER2 inhibitor. It is a strong CYP3A inhibitor, which makes CYP3A-substrate supportive drugs, including some antiemetics, sedatives and anti-infectives, difficult to combine with it. Capmatinib is kept because MET signalling participates in skeletal-muscle biology and has been studied in rhabdomyosarcoma. That is a biological association, not a labelled use. The interstitial-lung-disease warning, in a child born at 32 weeks, is why it stays limited.

Ponatinib 與 lapatinib 排除在這個孩子的用藥路徑之外。Ponatinib 的黑框涵蓋動脈阻塞、靜脈血栓、心衰竭與肝毒性。早產、生長失敗、肌肉萎縮和腎鈣質沉著不會讓這個警告比較容易接受。Lapatinib 因肝毒性黑框排除。標示還要求每一劑與餐食隔開至少一小時，因為食物，包含高脂餐，會提高暴露。規則是跟飯隔開，不是配高脂食物。對體重不增、進食困難的孩子，這個間隔並不友善。若實驗只問 knockdown 細胞核，六個仍可單藥測試。會帶去後續臨床討論的是 entrectinib 與 pralsetinib。Tucatinib 與 capmatinib 是備選。Ponatinib 與 lapatinib 不是。Ponatinib and lapatinib are excluded from a path toward use in this child. Ponatinib’s boxed warning covers arterial occlusion, venous thromboembolism, heart failure and hepatotoxicity. Prematurity, growth failure, muscle atrophy and nephrocalcinosis do not make that warning acceptable. Lapatinib is excluded for boxed hepatotoxicity. Its label requires each dose at least one hour before or one hour after a meal, because food, including a high-fat meal, increases exposure. The rule is separation from meals, not dosing with a high-fat meal, and that separation fits poorly with failure to thrive. If the only question is the knockdown nucleus, any of the six may still be tested singly. The names this filter would carry forward are entrectinib and pralsetinib. Tucatinib and capmatinib are backups. Ponatinib and lapatinib are not.

CMap 誘發 BUB1B mRNA 的老藥，能不能跟表 6 一起用CMap drugs that raise BUB1B mRNA, and whether they can be paired with Table 6

LINCS L1000 轉成 RNA-seq 的清單有 32 個 BUB1B 倍數大於 1 的簽名（results/bub1b_l1000_rnaseq_up.csv）。BUB1B 是 978 個地標基因之一，所以這些是探針實測，不是從其他基因推來的。每一列是一個細胞株、一個劑量、一個時間，連接分數大約 0.03。這不是蛋白穩定的測量，也不改表 6 的順序。A LINCS L1000-to-RNAseq list has 32 signatures with BUB1B fold change above 1 (results/bub1b_l1000_rnaseq_up.csv). BUB1B is one of the 978 landmark genes, so these values are measured rather than inferred. Each row is one cell line, one dose and one time, with connectivity about 0.03. This is not a protein-stability measurement, and it does not reorder Table 6.

倍數最高的是 CHK1 抑制劑 PF-477736，在 CJM 細胞 10 µM、24 小時為 1.94 倍。同一檔案裡還有 PLK 路徑的 HMN-214（1.27）與 Aurora 抑制劑 tozasertib（1.04），CMap 報告另列 barasertib。這些藥把細胞堆在 G2/M，BUB1B 才跟著上升。這正是表 6 已經排除的有絲分裂毒物。Ponatinib 也在檔案裡，XC.P907 細胞 0.12 µM、24 小時為 1.14 倍。它已經因黑框排除，mRNA 上升不會把它加回來。這些都不跟六個抗癌藥合併。The strongest rise is the CHK1 inhibitor PF-477736, fold change 1.94 at 10 µM for 24 h in CJM cells. HMN-214 (1.27) and the Aurora inhibitor tozasertib (1.04) are in the same file, and the CMap report adds barasertib. These drugs raise BUB1B while cells collect in G2/M. That is the mitotic-poison class already kept off Table 6. Ponatinib is also in the file, fold change 1.14 at 0.12 µM for 24 h in XC.P907 cells. The boxed warning already excludes it, and an mRNA rise does not put it back. None of these are partners for the six.

同一份上升清單裡、已上市且不是有絲分裂毒物的老藥是：risperidone 1.43、doxycycline 1.38、duloxetine 1.36、quetiapine 1.32、haloperidol 1.28、isradipine 1.26、bezafibrate 1.12、vorinostat 1.07。CMap 報告另引用 valproic acid，但它不在這 32 列裡，所以這裡不寫它的倍數。The marketed, non-mitotic names in the same up-list are risperidone 1.43, doxycycline 1.38, duloxetine 1.36, quetiapine 1.32, haloperidol 1.28, isradipine 1.26, bezafibrate 1.12 and vorinostat 1.07. The CMap report also cites valproic acid. That drug is not a row in the 32-line file, so no fold change is given here.

合併的想法是：多一點 BUB1B mRNA，加上一個面向剩餘激酶的分子，也許能提高檢查點蛋白的量。只有 p.Asn1002Lys 這個等位基因有機會受惠。p.Leu737Ter 是截斷。表 6 的藥並沒有被證明能穩定 BubR1，DynamicBind 沒有把 1002–1012 拉回野生型。所以這一對是細胞假說，不是處方。The pairing idea is more BUB1B mRNA plus a molecule aimed at the remaining kinase. Only the p.Asn1002Lys allele could benefit. p.Leu737Ter is a truncation. Table 6 drugs were not shown to stabilise BubR1: DynamicBind did not pull residues 1002–1012 back toward wild type. The pair is a cell hypothesis, not a regimen.

可以排進後續細胞實驗的，是一個已上市、非有絲分裂的誘導藥，加上第一梯隊的一個藥，而且兩邊都要先單藥讓 knockdown 細胞核移動。第一個要測的誘導藥是 doxycycline。它已上市，倍數在已核准藥裡偏高，而且不是強 CYP3A 抑制劑，不會踩到 pralsetinib 與 entrectinib 要避開中至強 CYP3A 抑制劑的規則。幼童仍有四環黴素對牙齒與骨骼的警告，旁邊又有生長失敗，所以它停在細胞候選，不是劑量。What can be staged next is one marketed, non-mitotic inducer plus one first-tier name, and only after each single moves the knockdown nucleus. Doxycycline is the first inducer to test. It is marketed, its fold change is among the higher approved-drug values, and it is not a strong CYP3A inhibitor, so it does not clash with the pralsetinib or entrectinib avoidance rules. In a young child the tetracycline warning on teeth and bone still applies beside existing growth failure, so it stays a cell candidate and not a dose.

抗精神病藥與 duloxetine 留下當簽名，不當合併對象。Quetiapine、risperidone、haloperidol 是 CYP3A 受質，也會延長 QT。Entrectinib 是中度 CYP3A 抑制劑，標示有 QT 警告。Tucatinib 是強 CYP3A 抑制劑。Duloxetine 有肝毒性警告，會跟這些激酶抑制劑已有的肝臟警告疊加。Isradipine 是 CYP3A 受質，抑制劑會提高它的暴露，它也會降血壓。Bezafibrate 在腎鈣質沉著旁邊需要腎臟謹慎。Vorinostat 與 valproic acid 是 HDAC 抑制劑。Vorinostat 本身是抗癌藥。Valproic acid 對幼童有肝毒性黑框。它們都不是 entrectinib 或 pralsetinib 的附加劑量。The antipsychotics and duloxetine stay signatures, not partners. Quetiapine, risperidone and haloperidol are CYP3A substrates and prolong the QT interval. Entrectinib is a moderate CYP3A inhibitor and has a QT warning. Tucatinib is a strong CYP3A inhibitor. Duloxetine carries a hepatotoxicity warning, which stacks with the liver warnings already on these kinase inhibitors. Isradipine is a CYP3A substrate, so a CYP3A inhibitor would raise its exposure, and it lowers blood pressure. Bezafibrate needs renal caution beside nephrocalcinosis. Vorinostat and valproic acid are HDAC inhibitors. Vorinostat is itself an anticancer drug. Valproic acid has a boxed hepatotoxicity warning that is worse in young children. Neither is an add-on dose for entrectinib or pralsetinib.

沒有親代樣本，所以反式是推論。沒有 RNA，所以無義媒介衰變是推論。沒有病人細胞。沒有做 N1002K knock-in 是篩選設計，用來讀 BubR1 失去，而不是補一個等位基因對照。SIFT、PolyPhen-2、CADD 是預測。Vina 不是解離常數。DynamicBind 親和力不是 Kd，而且這十四個柔性姿態並沒有把 1002–1012 拉回野生型；本表故意只採用它的親和力。泛素假說沒有西方墨點或質譜。HeLa 是一個 40× 視野。中藥單體沒有進入最後面板。本研究沒有做腫瘤體細胞定序。No parental sample, so trans is inferred. No RNA, so nonsense-mediated decay is inferred. No patient cells. The absent N1002K knock-in is the screening design, a readout of BubR1 loss rather than a missing allele-specific control. SIFT, PolyPhen-2 and CADD are predictors. Vina is not a dissociation constant. DynamicBind affinity is not a Kd, and these fourteen flexible poses did not pull residues 1002–1012 back toward wild type; this table uses the affinity on purpose. The ubiquitin idea has no blot or mass spectrum. The HeLa result is one 40× field. No herbal constituent entered the final panel. This study did not sequence the tumour.