Thalassemia, primarily including α- and β-thalassemia, is one of the most prevalent birth abnormalities in south China, (Lai et al., 2017). The imbalance between α- and β-globin chains determines the severity of clinical symptoms. Thalassemia’s clinical symptoms can vary from asymptomatic hemolytic anemia to fatal hemolytic anemia (Kattamis et al., 2022). The identification of high-risk partners through thalassemia carrier screening is an effective prevention of the birth of neonates with severe thalassemia. Hematological and biochemical tests are generally used first to identify individuals with phenotypic characteristics, followed by disease identification using genetic testing. Gap-polymerase chain reaction (gap-PCR), reverse dot-blot assay (RDB), and real-time PCR-based multicolor melting curve analysis (MMCA assay) are frequently used in China to detect common thalassemia genotypes, including four common α-thalassemia deletions, three common non-deletional α-thalassemia mutations, and seventeen known β-thalassemia mutations (Jiang et al., 2021a). However, approximately 2% of mutations cannot be detected using these techniques, making prenatal diagnosis of rare types of thalassemia mutations difficult (Li, 2017).

Molecular biological techniques have been developed rapidly in recent years, which significantly improved the diagnosis of rare thalassemia genotyping. Multiplex-ligation dependent probe amplification (MLPA) and Sanger sequencing are popular methods commonly used for detecting rare copy-number variations (CNV) and mutations. However, there are limitations associated with both methods. For example, Sanger sequencing produces high-quality sequences of only up to 1 kb (Krier et al., 2016). MLPA only identifies changes within the probes’ target sequence. Copy number neutral inversions or translocations remain undetected using both methods, making inversions or translocations with copy number neutrality unnoticed.

Due to high homology and high GC-rich regions of the HBA2 and HBA1 genes, the accuracy of NGS (next-generation sequencing) methods based on short reads is not optimal for globin cluster variants (Xu L. et al., 2020). Third-generation sequencing (TGS) has been used in thalassemia genetic testing recently. This method is based on Pacific Bioscience (PacBio) single-molecule real-time sequencing (SMRT) and has been proven to be able to detect structure variations (SVs), especially the ability to effectively and accurately distinguish complex rearrangement due to its advantages of long reads and single-molecule resolution. In clinical settings, to reduce errors in raw subhead coverage and decrease costs, SMRT is commonly performed following multiplex long-molecule PCR. However, it remains challenging to provide more precise information on positions for unknown deletions or duplications that occur outside of the detection range. In this study, we conducted a thorough analysis of thalassemia alleles for a better prenatal diagnosis and counseling of thalassemia carriers.

Thalassemia is the most common birth defect in Guangdong province, China. It has been clarified to be very useful to perform prenatal diagnosis in couples who are carriers of the same type of thalassemia. Traditional methods for detecting population-specific targeted mutation are widely used in prenatal diagnosis. With the development of molecular biology methods, more and more at-risk couples with rare thalassemia variants have been offered prenatal diagnosis. Sanger sequencing and MLPA have been used in the prenatal diagnosis of rare thalassemia. The former is used for point mutation and the latter is applicable for CNV detection. Although targeted next-generation sequencing was reported to detect an additional 35 at-risk couples from 2522 subjects compared to traditional methods, it was rarely used in prenatal diagnosis of thalassemia due to high cost (Shang et al., 2017). A long-read sequencing-based approach was first used in prenatal diagnosis of thalassemia in 2022, which additionally identified α-globin triplicates in two fetuses with the heterozygous HBB c.316-197C>T variant (Luo et al., 2022a). The prenatal diagnosis rate of rare types of thalassemia was only 0.12% from 2011 to 2019 while it was 3.26% from 2020 to 2022, mainly because of the application of new genetic analysis technologies such as MLPA (Xu L. L. et al., 2020). The gene spectrum has also been changed especially for β-thalassemia. For example, Chinese ^Gγ(^Aγδβ)⁰ thalassemia/SEA-HPFH combined with point mutations in the HBB gene is the most common and rare point mutations in the HBB gene are a main cause for prenatal diagnosis in at-risk couples with a history of having children with severe β thalassemia syndrome (Xu L. L. et al., 2020). We found that about 1.13% of the couples were heterozygous for Chinese ^Gγ(^Aγδβ)⁰ thalassemia and SEA-HPFH in Guangzhou city. The most common characteristic of carriers with Chinese ^Gγ(^Aγδβ)⁰ thalassemia/SEA-HPFH is a high HbF level. Gap-PCR is a cost-effective method to detect those two types of thalassemia, especially in basic-level hospitals, thus we suggested traditional methods should cover those two genotypes (Jiang et al., 2020).

A combination of different methods was performed in this study. We additionally detected the complex genotypes of thalassemia which were different from previous results. The most representative one was a fetus with Chinese ^Gγ(^Aγδβ)⁰ deletion combined with HBG2-HBG1 fusion using MLPA. His father was the carrier of Chinese ^Gγ(^Aγδβ)⁰ deletion while his mother was the--^SEA/αα carrier coexisted with β^{CD41-42(−CTTT)}/β^N based on the result from another hospital. The HBG2-HBG1 fusion was not detected previously because the deletion (hg38, Chr11: 5226187-5231089) does not locate at the common Chinese ^Gγ(^Aγδβ)⁰ deletion locations (hg38 chr11:5169921-5248825). There is about 520 bp between those two deletions which does not affect primers for Chinese ^Gγ(^Aγδβ)⁰ deletion to bind. Using MLPA, we were able to simultaneously detect those two deletions. It is the first time that the HBG2-HBG1 fusion was reported and found to cause the high HbF level by our team. We also found that the HbF was not presented in the proband with the same fusion described by Luo et al. (2022b). In this family, the HbF value was only 1.1% for the mother. The hematological phenotype of the woman and her 6-month-old baby described before was similar to that of ß-thalassemia minor genotypes, suggesting that HBG1-HBG2 4,924 bp deletion coexisted with HBB gene mutation does not cause the severe phenotype. The fetus was diagnosed with γ thalassemia because he had only one normal ^Gγ globin gene and an HBG2-HBG1 fusion gene. The γ globin gene is mainly expressed during the fetal period, which includes the ^Gγ and ^Aγ globin genes. The ^Gγ^/Aγ ratio is about 7:3 in normal conditions, making the influence of the deletion in this fetus not severe (Nemati et al., 2010).

The importance of the combination of different methods is that we could identify different genotypes even if we obtain the same result based on one method. We have used MLPA to find a novel 8.2 kb heterozygous deletion in the β-globin gene cluster, but it was difficult to obtain accurate breakpoints. SMRT sequencing could accurately estimate the deletion range. The deletion only influenced the β-globin gene and its close upstream and downstream areas. Typically, there is a strong rivalry or competition for control between the promoter region of the β gene and the promoter regions of the γ and δ genes. The β gene promoter, responsible for the production of β globin chains, and the promoters of the γ and δ genes, responsible for the production of γ and δ globin chains, respectively, are vying for transcription factors to activate their respective genes. This competition determines the relative amounts of β, γ, and δ globin chains produced in a cell or tissue. The competition between these promoters can have important implications for the synthesis of hemoglobin, as different combinations of globin chains give rise to different types of hemoglobin. Thus, the deletion of the β-globin gene including the promoter may make the remaining γ- and δ-globin gene promoters react more readily with the enhancer elements and may result in higher levels of Hb A2. There were no significantly homologous sequences in the deleted region compared to the normal 5′and 3′sequences, suggesting that the deletion was caused by a non-homologous recombination event. There were AT-rich stretches around the 3′breakpoint, which may be significant in the recombination. A similar situation was observed in Taiwanese β-thalassemia and the case with a novel 7.2 kb deletion (Chr11:5222800-5230034, hg38) located in the HBB gene (Zhong et al., 2022). According to the phenotype of the proband and that of those two deletional β-thalassemia carriers, the deletion may cause β⁰ thalassemia. Interestingly, the result of MLPA for the novel 7.2 kb deletion was the same as the one we had described (Zhong et al., 2022). It has been shown that SMRT sequencing played an important role in thalassemia breakpoint detection. We also found another deletion which included up-, down-stream, and the whole HBB gene, which was recently reported in a Chinese family. The Hb A2 level and Hb F level were significantly high in patients with the two deletions, which was also observed in Taiwan deletion. Topfer et al. (2022) used CRISPR gene editing to delete elements within the HBB gene promoter and found that it could reduce HBB promoter activity, resulting in elevated fetal globin expression. However, whether the deletions in those regions could cause high HbA2 remained unclear. As rearrangement is very common in the α-globin gene cluster due to three homologous X-, Y-, and Z-box segments, it was difficult to discriminate the case with rare micro gene conversion of HBA2 gene combined with--SEA deletion from the Hb Bart’s hydrops fetus using Gap-PCR. Gap-PCR was widely used in prenatal diagnosis and we previously reported a woman with this genotype using this method. The patient had classical mild α-thalassemia traits, suggesting the conversation would not influence phenotype. -α3.7 is the most common silent α-thalassemia and ααα^anti−4.2 is the most common triplicated allele in China (Jiang et al., 2021a; Luo X. et al., 2021). In this study, we also obtained the same result of MLPA in those two rare micro gene conversions of the HBA2 gene. Previous studies have shown that -α3.7 is the most common silent α-thalassemia and ααα^anti−4.2 is the most common triplicated allele in China (Yin et al., 2014; Luo X. et al., 2021; Long and Liu, 2021). Based on our clinical observation and analytic results, we hypothesize that the genotype of one chromosome is -α3.7 and the other is ααα^anti−4.2, and two Z homologous segments are unequally crossover and recombined to produce those two rare conversations (Supplementary Figure S2). Although the two rare micro gene conversions of the HBA2 gene have been reported by our team, this is the first time to describe their phenotypes in family numbers. Our findings could give more information for prenatal diagnosis of thalassemia, especially rare types.

Duplications in the α globin gene combined with mutations in the β globin gene were another common reason for prenatal diagnosis of thalassemia (Shooter et al., 2015; Clark et al., 2018). MLPA, SMRT, and next-generation sequencing are useful methods for identifying large duplications in hemoglobinopathies (Jiang et al., 2022). It is difficult for MLPA to investigate chromosomal regions larger than 130 kb and determine the breakpoint. Compared to MLPA, SMRT, and next-generation sequencing can give precise information for the location of structural change in α and β globin genes simultaneously, except that they cannot capture the whole repetitive sequence (Ardui et al., 2018). Sequencing cost will also increase when more primer pairs are used in SMRT sequencing, which limits the detection range of the SMRT method to less than 10 kb (Luo S. et al., 2021). The situation also exists in targeted next-generation sequencing although the detection range is more than 100 kb. For the two rare cases reported in the study, more information about breakpoints could be obtained using targeted next-generation sequencing unless the breakpoint occurs in this sub-telomeric region of chromosome 16. The situation was also observed when we detected large deletions in α or β globin gene cluster.

--^THAI deletion was the most common deletion in prenatal diagnosis for rare α thalassemia in our study. MMAC assay includes this deletion but the result should be identified using another method. Another rare 11.2 kb deletion of the α-globin gene cluster was also common in prenatal diagnosis. It was first described in a case of Hb H disease by our team (Jiang et al., 2021b). Using Target Capture-Seq, the exact breakpoint location was detected and verified by Sanger sequencing using a special primer. The clinical phenotype of probands supplied by the 11.2 kb deletion could cause α⁰-thalassemia. Based on the two families, the phenotype of the rare 11.2 kb deletion carrier was similar to that of the–^SEA carrier. The result of follow-up for the proband with α^11.1kb/-α^4.2 supported that this rare deletion combined with other α⁰-thalassemia would not cause the severe clinical phenotype.

In conclusion, with the development of new sequencing technologies, more rare genotypes of thalassemia were detected, but this also provides a challenge for genetic counseling and prenatal diagnosis. In our study, we described the rare genotypes and clinical phenotypes in prenatal diagnosis. Chinese (^Aγδβ)⁰ and SEA-HPFH should be included in the detection range of conventional methods. It is important to identify the genotypes of parents before prenatal diagnosis, especially in cases with phenotypes not matching genotypes. MLPA for α globin cluster should be performed on cases with severe β− thalassemia who only carry a mutation in the HBB gene. Compared to other methods, SMRT sequencing has incomparable advantages to find complicated genotypes of thalassemia.