Development of SPF PCR primers.
Complete or partial HPV sequences were obtained from GenBank and used for alignment of L1 region sequences. As described earlier (17
), the L1 region is relatively well conserved, and several general PCR primers in this part of the HPV genome were developed (Fig. ). Recently, we have selected a novel set of PCR primers in this region which amplify a fragment of only 65 bp (17
). These primers, designated SPF, allowed extremely sensitive amplification from a broad spectrum of HPV genotypes. The present study aimed at using the sequence variation within the amplified fragment of 65 bp for identification of specific HPV genotypes. The SPF primers flank an interprimer region of 22 bp. The alignment of the complete 65-bp fragments amplified by SPF primers from the L1 regions of different mucosal HPV genotypes is shown in Fig. . Apparently, all HPV genotypes except genotypes 68 and 73 have a unique SPF interprimer sequence.
FIG. 2 Alignment of 65-bp nucleotide sequences from the L1 regions of a total of 39 HPV genotypes. The target sequences for the SPF primers are boxed and flank the 22-bp interprimer region, as indicated. Positions are according to the HPV-16 sequence PPH16 (GenBank (more ...) Inter- and intratypic sequence conservation in the SPF 22-bp interprimer region.
To investigate whether the observed sequence variation among and between HPV genotypes is consistent, a total of 134 sequences from 31 mucosal HPV genotypes, obtained from GenBank, were analyzed (Table ). For each genotype, the full-length genomic sequence was used as a reference (see Materials and Methods). In 8 (5.9%) of the 134 sequences studied, the interprimer sequences were not completely conserved compared to their respective reference sequences. Four of the nine HPV-18 sequences showed identical single-base-pair mismatches to the reference sequence, and this variant was provisionally designated HPV-18var. One of the two HPV-57 sequences contained a single mismatch, and this sequence had been formally classified as HPV-57b (33
). Thus, in these five cases the correct HPV type can be recognized by analysis of the 22-bp sequences. In contrast, one of the HPV-11 sequences showed five mismatches compared to the reference sequence. Among 28 HPV-16 sequences, one sequence contained a single mismatch to the reference sequence. One of the HPV-54 sequences showed two mismatches to the reference. Taken together, analysis of the 22-bp sequences resulted in identification of the correct HPV in 131 (97.8%) of the 134 cases.
TABLE 1 Conservation of the 22-bp SPF interprimer sequences among mucosal HPV genotypes for 134 HPV sequences obtained fromGenBanka
Since the number of L1 region sequences available in GenBank was limited for several HPV genotypes, a total of 104 clinical isolates were also analyzed. HPV DNA was amplified from these samples by the MY 09/11 as well as the SPF primers. Identification of HPV genotypes was based on sequence analysis of the MY 09/11 amplimer. SPF amplimers were also analyzed, and the HPV genotypes were deduced from the 22-bp SPF interprimer sequences. The results are shown in Table . Among the 104 isolates studied, MY 09/11- and SPF-based typing results were initially discordant in 7 cases (5.1%). These discordant cases were further analyzed. Among 19 isolates identified as HPV-16 by the MY 09/11 sequence, SPF analysis showed the presence of HPV-31 in 1 isolate. Type-specific PCR revealed the presence of both HPV-16 and -31 in this isolate. Similarly, in one sample containing HPV-18, SPF PCR detected HPV-16, and the presence of both HPV-16 and -18 was confirmed by type-specific PCR. In the isolate containing HPV-51, sequence analysis of the SPF fragment revealed a single mismatch to the HPV-51 reference sequence. Two of the five samples with HPV-56 yielded discordant results. In one case, the 22-bp sequence showed a single mismatch to the HPV-56 reference sequence, whereas the other case showed the presence of HPV-45. One case of HPV-58 was mistyped as HPV-56 by the SPF system. In the sample containing HPV-73, the SPF system detected HPV-53. Since type-specific primers are not available for HPV-51, -53, -56, -58, and -73, not all discordant results could be analyzed by type-specific PCR. Thus, the HPV types identified from the 22-bp SPF interprimer sequences were initially concordant with MY 09/11 sequences in 99 (95.2%) of the 104 cases studied. Several discordant results were suspected to be due to the presence of multiple HPV types, and this was further analyzed by reverse hybridization (see below). These results show that sequence variation of the SPF amplimer can be used to identify a broad range of HPV genotypes.
Development of a reverse hybridization LiPA.
To identify HPV genotypes by hybridization, specific probes were deduced from the SPF sequence alignments (Fig. ) and used to develop the INNO-LiPA HPV prototype research genotyping assay. Since HPV genotypes often differ by only a single nucleotide in the 22-bp interprimer sequences, well-controlled hybridization conditions are necessary. Therefore, a reverse hybridization LiPA was developed (27
), allowing the simultaneous identification of multiple HPV genotypes in a single hybridization step. All probes were optimized to ensure that hybridization occurs only between completely matching sequences. The outline and representative examples of the HPV LiPA are shown in Fig. .
FIG. 3 Outline and representative examples of the INNO-LiPA HPV genotyping assay. The positions of the control line and the 28 specific probes for each of the 25 HPV genotypes are shown at the left. The number above each strip indicates the HPV genotype for (more ...)
Most of the HPV genotypes are recognized by hybridization to a single type-specific probe. However, a number of HPV genotypes, i.e., 6, 31, 33, 40, 45, 56, 58, 68, and 74, hybridize to more than one probe and can be directly recognized by a specific hybridization pattern on the strip.
To assess the efficacy and reliability of the LiPA, HPV sequences were amplified by SPF primers from a total of 34 plasmids containing complete or partial HPV genomic sequences. The relevant part of the L1 region was amplified from these plasmids and analyzed by direct sequencing. SPF LiPA yielded the expected HPV genotyping results in all cases, and these were completely concordant with sequence data, indicating the high specificity of the reverse hybridization assay.
To determine the efficacy of the SPF system to detect multiple HPV genotypes, mixtures of HPV-11 and HPV-18 target DNAs were tested. A fixed quantity of HPV-11 was combined with increasing concentrations of HPV-18. Conversely, a fixed quantity of HPV-18 DNA was combined with various concentrations of HPV-11. All mixtures were amplified by the SPF primers and tested in the LiPA. The results indicated that the SPF system permits detection of two different HPV genotypes, even if one type is present in a 1,000-fold excess over the other type (data not shown). In contrast, it appears that sequence analysis permitted type-specific detection of both types up to only a 10-fold excess of one genotype over the other.
The amplification target of the SPF primers is located within the target region for the MY 09/11 primer set. Therefore, SPF as well as MY 09/11 amplimers can be used for analysis in the LiPA, provided that PCR primers are biotinylated. SPF and MY 09/11 amplimers from the 104 clinical samples of patient group 1 had already been analyzed by direct sequencing as described above and shown in Table . Subsequently, these amplimers were also tested by LiPA. In one HPV-16 isolate, the presence of HPV-16 and -31, as also determined by type-specific PCR, was confirmed, indicating infection with multiple HPV types in this case. Similarly, the presence of both HPV-16 and -18 in the single discordant case of HPV-18 was confirmed. Since the single mismatch was not included in the type-specific probe for these genotypes on the LiPA strip, the HPV-51 isolate and one of the HPV-56 isolates that showed a mismatch to their reference sequences were both correctly typed by the LiPA. In the remaining discordant case of HPV-56, a mixture of HPV-45, -52, and -56 was found. By using the SPF system, one of the HPV-58 isolates was identified as HPV-56 by sequencing as well as by LiPA. The sequence of the MY 09/11 amplimer classified this isolate as HPV-58, but the 22-bp SPF interprimer region in this amplimer was completely homologous to that of HPV-56 (Fig. ). In the 22-bp SPF interprimer region, the HPV-58 genome differs by only two nucleotides from HPV-56. Therefore, this isolate can be considered as being mistyped by SPF due to intratypic sequence variation of HPV-58. Finally, LiPA analysis of the sample containing HPV-73 revealed the presence of a mixture of HPV-53 and HPV-68 or -73.
Taken together, and including the cases with multiple HPV genotypes, the LiPA results were in agreement with the sequence analysis of the SPF and MY 09/11 fragments, as well as type-specific PCR, in 103 (99%) of the 104 cases.
Evaluation of the HPV LiPA with clinical samples.
To assess the performance of the HPV LiPA system, various groups of clinical specimens were investigated. All clinical samples yielded a β-globin-specific amplimer, confirming the presence of amplifiable DNA. A total of 488 cervical scrapes from patient group 2, classified as having normal cytology or ASCUS, were tested by SPF PCR, as well as with the GP5+/6+ primers. SPF PCR detected HPV DNA in 117 (24%), whereas GP5+/6+ detected HPV-DNA in only 77 (15.7%), of the cases. SPF PCR was repeated for cases that were exclusively positive by SPF to confirm the presence of HPV DNA. The GP5+/6+-positive samples were all positive by SPF PCR. To confirm the specificity of the SPF primer set, the resulting amplimers were subjected to sequence analysis, and the results are shown in Table . Based on 100% identity to reference sequences, the 22-bp sequences could be assigned to known HPV genotypes in 93 (79.4%) of the SPF-positive cases, and multiple HPV types were detected in 12 samples (10.2%). GP5+/6+ PCR did not detect HPV DNA in 40 (34.2%) of the 117 SPF-positive cases, and these GP5+/6+-negative cases were distributed over multiple HPV genotypes, suggesting a higher sensitivity of SPF over a broad range of HPV genotypes.
TABLE 3 Detection of different HPV genotypes by SPF and GP5+/6+ PCR among 488 cervical scrapes from group 2 (1997) and 278 cervical scrapes from group 3(1998)
A total of 278 cervical scrapes in patient group 3, also classified as having normal cytology or ASCUS, were investigated by SPF and GP5+/6+ PCR. These results are also shown in Table . SPF PCR yielded positive results in 70 samples (25.2%), whereas the use of GP5+/6+ resulted in only 52 positive cases (18.7%). SPF LiPA detected multiple HPV types in 10 (14.2%) of the SPF-positive samples. The cases that remained undetected by GP5+/6+ but were SPF positive were distributed over various HPV types. In 8 (11.4%) of the 70 SPF-positive samples, the HPV type could not be assigned.
Altogether, a total of 766 cervical scrapes (groups 2 and 3), were analyzed, of which 697 (90.1%) had normal cytology and 69 (8.9%) were classified as ASCUS. Among the cases with normal cytology, SPF PCR detected 157 (22.5%), whereas GP5+/6+ detected only 101 (14.4%), HPV-positive samples, and this difference was highly significant (P < 0.001). Among the 69 cases classified as ASCUS, SPF primers detected more HPV-positive samples (30/69 = 43.5%) than the GP5+/6+ primers (26/69 = 37.7%), but this difference was not statistically significant.
Among all SPF-positive samples, 8 (53.3%) of 15 samples containing HPV-31 were GP5+/6+ negative. Similarly, 6 (50%) of the 12 HPV-66 cases remained undetected by GP5+/6+ (Table ). These findings indicate a significantly lower sensitivity of the latter primer set for these particular HPV types. Sequences from 32 (17.1%) of the 187 SPF-positive samples could not be assigned to any known HPV genotype. Of these 32 samples 26 were tested with the MY 09/11 primers, and 8 were HPV positive. All interprimer sequences were exactly 22 bp in length and were similar to HPV sequences, but they showed between one and five mismatches to any known HPV genotype. Among the 155 samples containing a known HPV genotype, 32 (20.6%) were infected by an HPV with a low-risk genotype (HPV-6, -11, -42, -44, -53, -54, -55, -61, -62, -67, -70, -74, or -MM7).
The fourth group comprised 304 selected cervical smears with mild to moderate (n = 151) or severe (n = 153) dyskaryosis. A total of 299 (98.4%) of the 304 samples were positive by SPF PCR. HPV genotypes were determined by SPF LiPA, and the results are shown in Table . Of these, 147 (97.3%) of the 151 cases with mild or moderate dyskaryosis were SPF positive, and 95 of these contained a single HPV genotype. In 2 (1.4%) of the 147 SPF-positive samples, the sequences could not be assigned to a known HPV type, and these will require further characterization. High-risk HPV genotypes (HPV-16, -18, -31, -33, -35, -45, -51, -52, -56, -58, and -66) were present in 95.2% of the cases. Similarly, 152 (99.3%) of the 153 scrapes diagnosed with severe dyskaryosis were HPV positive. HPV genotypes could be assigned in 151 (99.3%) of these samples, and all were classified as high-risk types.
TABLE 4 Detection of HPV genotypes among 304 cervical smears with mild or moderate dyskaryosis or severe dyskaryosis (group4)
Altogether, the SPF sequences could not be assigned to a known HPV genotype in only 3 (1%) of the 304 cases. A total of 105 (34.5%) of the 304 samples contained more than one HPV genotype, and in 103 (98.1%) of these, at least one high-risk genotype was detected. In 21 (6.9%) of the samples, three HPV genotypes were detected, and in 5 (1.6%) cases, four different HPV types were observed.
Group 5 contained 180 formalin-fixed, paraffin-embedded cervical carcinoma samples, comprising 51 adenocarcinomas and 129 squamous cell carcinomas. SPF PCR yielded positive results in all cases. HPV genotypes were identified by sequence analysis and LiPA, and only high-risk HPV types were found, as shown in Table . HPV-16 and -18 were found in more than 70% of both groups. The relative prevalence of HPV-18 (17.6%) among the adenocarcinomas was higher than that among the squamous cell carcinomas (6.2%), and this difference was statistically significant (P = 0.027). The number of samples containing other high-risk HPV types was too small to analyze differences between the two groups of carcinomas.
TABLE 5 HPV genotypes among 180 cervical carcinomas as determined by sequence and LiPA analysis of SPF PCRfragments