Fluorescence in situ hybridization (FISH) break-apart probes are indispensable for detecting gene rearrangements in malignant cells. These probes utilize dual-color fluorophores to assess and visualize chromosomal integrity, facilitating the identification of structural aberrations such as translocations and inversions [1–4]. Typically, the probe hybridizes to the correct chromosomal location, generating the expected signals: intact loci yield fused signals, while translocations yield split signals [5–7]. However, tumors with complex genetic alterations frequently generate atypical signals, particularly isolated signals (characterized by the loss of signal from one probe) that challenge diagnostic paradigms [8–10]. Notably, in clinical practice, isolated signals are far rarer than classic split signals, which are commonly observed in translocation-positive malignancies. For classic split signals, the criterion for defining FISH positivity is well-established, usually requiring 15% or more of cells to display split signals [11–13]; in contrast, few reports specify the threshold proportion of cells with isolated signals that indicates FISH positivity, further complicating the standardized interpretation of such atypical signals [14]. Isolated signals are observed in only 0.1%–28.9% of all malignant cases undergoing break-apart FISH testing [8, 12, 13, 15–21]. Most large-cohort studies indicated rates below 10% [12, 19–21]. This rarity, together with the diagnostic ambiguity of isolated signals, further complicates accurate interpretation, as clinical laboratories often have limited experience with such infrequent signal patterns.

Isolated signals have been reported in break-apart probes for multiple genes, including SS18 [5, 16, 17, 22], BCL6 [18], ALK [12, 19, 20], ROS1 [23, 24], EWSR1 [13, 25], DDIT3 [8], FUS [8, 26], USP6 [8], CBFB [21], MLL [27] and TFE3 [28], with varying interpretations across different studies. Notably, most existing hypotheses and inferences regarding the formation of these isolated signals have focused on the target genes themselves, primarily attributing their occurrence to deletions or translocations. For instance, isolated signals in the SS18 break-apart probe, usually associated with loss of either the 5′or 3′signal, typically arise from specific unbalanced rearrangements [22], partial deletions of the SS18 gene [17], or deletions of the SS18-SSX fusion gene [5]. Isolated 3′signals for EWSR1 and TFE3, as well as isolated 5′signals for CBFB, are reported to arise from unbalanced rearrangements of the corresponding genes [13, 21, 28]. Isolated 5′signals in ALK may arise from deletions of the ALK 3′ region [19, 20]. Deletion of the 5′signal for FUS was thought to arise from supernumerary ring chromosomes [26]. For break-apart probes of BCL6, ROS1, DDIT3 and USP6, isolated signals have occasionally been reported [8, 18, 23, 24], suggesting that further research is indicated. Although various hypotheses and inferences have been proposed regarding isolated signals, the exact mechanisms underlying them remain uncharacterized, with no consensus on their biological significance or clinical interpretation.

A recent large-cohort study demonstrated that break-apart probes of MYC, BCL2, BCL6 can miss cryptic rearrangements due to small chromosomal insertions or inversions, yet their work did not address the distinct atypical pattern of isolated signals, whose genomic basis remains unclear [29].

In this study, six cases of lymphoma and myxoid liposarcoma with isolated FISH signals were investigated using genome resequencing to characterize the underlying genetic alterations. Additionally, the expression of the corresponding proteins was assessed. These investigations aimed to elucidate the mechanism of formation and diagnostic significance of isolated signals in FISH break-apart probe assays, which may ultimately aid in the development of evidence-based clinical guidelines.

The genetic heterogeneity of tumor cells often gives rise to atypical FISH signals, especially isolated signals, using break-apart probes to perform gene translocation analysis. In this study, we identified six cases exhibiting isolated FISH signals: three cases showed isolated 5'/3′signals for BCL2, MYC, or BCL6, while the remaining three cases displayed isolated 3′signals for DDIT3.

WGR performed on all six samples with isolated signals for BCL2, MYC, BCL6, or DDIT3 revealed complex genomic rearrangements, with no case in this cohort showing isolated signals without concurrent genomic rearrangements. These changes included novel gene fusions (e.g., MAP2K1) involving BCL2, intergenic rearrangements (e.g., ELK2AP/MIR4507) affecting MYC, novel BCL6 fusions (e.g., SNHG29), and rare EWSR1-DDIT3 fusions (alongside the two classic FUS-DDIT3 fusion). Distinct from previously reported fusion partners of BCL2, MYC, and BCL6 [29–32], this is the first report, to our knowledge, of these novel partners for the three genes.

More significantly, complex genetic alterations, including deletions, inversions, and multiple SVs (inter- and intra-chromosomal translocations), were identified in the binding regions of FISH break-apart probes for the six cases with isolated signals by FISH analysis, with no deletions or inversions detected in the target genes (BCL2, MYC, BCL6, DDIT3) or their translocation partners [33]. For instance, deletions and inversions were detected in the 5′probe-binding region of BCL2 in our case, along with an inversion in its 3′probe-binding region. Deletions and inversions were also identified in both the 5′and 3′probe-binding regions of MYC and BCL6. Additionally, WGR analysis revealed complex genetic alterations in the 5′probe-binding regions of DDIT3, including large deletions, inversions, and multiple SVs. By contrast, in control tumor samples with classic split signals, the probe-binding regions of BCL2, MYC and DDIT3 harbored only multiple SVs, whereas those of BCL6 exhibited limited deletions alongside multiple SVs. This stark contrast suggests that extensive deletions or inversions in probe-binding regions are unique to tumors with isolated signal in the cases analyzed, and likely linked to their formation.

To interpret this distinction, we first wish to highlight the design principle of commercially available break-apart probes: probes targeting BCL2, MYC, BCL6, and DDIT3 all adopt a dual-fluorophore strategy, labeling flanking sequences of the target gene (rather than the gene’s coding region itself) to assess chromosomal integrity (Figures 3A, 4A, 5A, 6A). Considering this mechanism and our WGR results, we propose that isolated signals arise due to complex genetic alterations in the probe-binding regions, rather than the target genes; deletions or inversions disrupt the recognition and binding of fluorophore-labeled probes, leading to the loss of one signal (red or green) and thus isolated signals. This mechanism is further supported by previous studies: Pacheco et al. [25] reported a deletion encompassing the SMARCB1 locus on chromosome 22 in an atypical teratoid rhabdoid tumor case with isolated 3′EWSR1 signals; Ordulu et al. [34] identified microdeletions in the telomeric and centromeric regions of 7p at the JAZF1 locus in a low-grade endometrial stromal sarcoma case with 1F JAZF1 signals; and Yang et al. [21] detected 16q inversions in eight acute myeloid leukemia cases with isolated 5′CBFB signals.

Zeng et al. [29] large-cohort study in DLBCL focused on MYC, BCL2, and BCL6 and identified “FISH-cryptic rearrangements” (no observable signal abnormality, only detectable by NGS) caused by small insertions or inversions. In contrast, our study characterizes isolated signals as a distinct atypical pattern driven by deletions or inversions in probe-binding regions that highlights a unique genetic mechanism underlying this specific FISH signal anomaly.

Notably, in the cases analyzed, since all break-apart probes (targeting BCL2, MYC, BCL6, and DDIT3) share the core design principle of labeling target gene flanking regions, the identified cause of isolated signals was consistent across these probes (i.e., probe-binding region alterations rather than coding region-specific mechanisms). This observation provides a preliminary basis for exploring similar interpretations of isolated signals in other break-apart probe-targeted genes with analogous flanking sequence labeling principles, but generalizing interpretations of isolated signals across break-apart probe targeting genes with similar flanking sequence labeling principles requires further validation in larger, more homogeneous cohorts.

Immunohistochemical analysis revealed high expression of BCL2, c-MYC, and BCL6 in cases with isolated 5'/3′signals for the respective genes. WGR confirmed the presence of translocations involving these genes in all of the tumors with isolated signals, while no target gene amplification was detected. Tay et al. [17] detected the expression of SS18-SSX proteins in synovial sarcoma cases with isolated 5′SS18 FISH signals, and next-generation sequencing results confirmed the occurrence of SS18-SSX fusions. However, in cases without fusion proteins, SS18 translocation was not detected. Li et al. [19] detected ALK protein expression in non-small cell lung cancer cases with isolated 5′ALK signals and in those with isolated or attenuated 3′signals. Next-generation sequencing confirmed the occurrence of ALK fusions in these cases. Zeng et al. [29] revealed that all cases with FISH-cryptic MYC, BCL2, or BCL6 rearrangements were positive for the corresponding proteins. In this study, the coexistence of high BCL2/c-MYC/BCL6 protein expression, confirmed gene translocations, and absence of target gene amplification strongly suggests that the observed upregulation of these proteins is likely driven by the translocations involving their respective genes, rather than by gene amplification events.

It is important to acknowledge this study’s limitations. First, only six isolated signal cases were analyzed, including heterogeneous malignancies (e.g., FL, BL, DLBCL, MLPS). The small sample size and inherent tumor type heterogeneity significantly weakens statistical power and restrict the generalizability of our findings. The proposed mechanism of isolated signal formation, therefore, may not apply to all tumor types, all break-apart probe-targeted genes, or larger, more homogeneous cohorts. Second, the single-center retrospective design of this study introduces additional potential biases. Reliance on archived samples meant that only specimens with sufficient tissue integrity and high-quality DNA extraction for WGR were included, while samples with severe DNA degradation, insufficient tissue volume, or poor preservation were excluded. This selection bias may have overrepresented cases with clear and detectable genetic alterations in probe-binding regions, potentially skewing the correlation between these alterations and isolated signal formation. Furthermore, tissue quality directly impacts WGR data accuracy: degraded DNA can lead to incomplete genomic coverage, missed detection of subtle deletions or inversions in probe-binding regions, and inaccurate identification of translocation breakpoints, all of which may compromise the reliability of our mechanistic inferences. Notably, the long-term stored archived samples also resulted in poor RNA quality, precluding RNA-based orthogonal confirmation of fusion transcripts. Additionally, the single-center setting limits the diversity of tumor subtypes and clinical backgrounds, further constraining the generalizability of our conclusions. As such, the present study should be explicitly considered a preliminary exploration of the genetic mechanism underlying isolated FISH signals. The conclusions drawn are tentative and require validation through future prospective studies featuring larger, well-stratified cohorts (with homogeneous tumor types and increased sample sizes) and longitudinal sampling to determine the broader applicability of the proposed mechanism. Additionally, no long-term follow-up data on treatment response and prognosis were collected, which precluded the assessment of the clinical implications of isolated FISH signals. Future studies with extended clinical follow-up are warranted to clarify the prognostic and therapeutic relevance of these atypical signals.

Isolated signals detected by FISH break-apart probes for BCL2, MYC, BCL6, and DDIT3 may be attributed to deletions or inversions in the probe-binding sequences for these genes (not the target genes themselves). Notably, in cases with isolated BCL2, MYC, or BCL6 signals, our data showed an association between translocations involving these genes and increased expression of their encoded proteins.